MIT News

MIT engineers find a way to deliver drugs directly to the esophagus

Anne Trafton | MIT News — Fri, 12 Jun 2026 05:00:00 -0400

There are few treatment options available for people with disorders of the esophagus. Delivering drugs directly to this part of the body is difficult, so patients are usually treated with systemic drugs, which can have unwanted side effects.

To overcome that challenge, MIT engineers developed a gel-like oral drug formulation that can coat the mucosal lining of the esophagus after being swallowed, allowing drugs to pass through the tissue.

The formulation, which includes a hydrogel and other key ingredients that promote rapid drug absorption, could be used to deliver antibodies including infliximab, used to treat a number of autoimmune diseases, or other types of antibodies or small-molecule drugs.

“There are many people with esophageal disease, and if you look at drugs for these conditions, they’re very limited in their ability to target this part of the body and it’s very difficult to develop them. We hope this platform will make it easier to develop systems that can help patients suffering from these conditions,” says Giovanni Traverso, an associate professor of mechanical engineering at MIT, a gastroenterologist at Brigham and Women’s Hospital, and an associate member of the Broad Institute of MIT and Harvard.

Traverso is the senior author of the new study, which appears today in Nature Biomedical Engineering. Former MIT postdoc Christina Karavasili, now an assistant professor at Aristotle University of Thessaloniki in Greece, is the paper’s lead author.

Direct delivery

One of the most common disorders of the esophagus is eosinophilic esophagitis, a type of inflammation that is caused by food allergies and leads the esophagus to close up, making it impossible to swallow food. Crohn’s disease can also cause inflammation of the esophagus.

These disorders are usually treated with systemic drugs, including infliximab, an antibody that neutralizes an inflammatory protein called tumor necrosis factor alpha (TNF-alpha). However, this drug is an immunosuppressant that can lead to a higher risk for infections and other health problems.

Delivering the drug directly to the esophageal tissue could reduce those side effects, but this is inherently challenging because drugs taken orally pass through the esophagus so quickly. Adding to the difficulty, the esophagus is lined by a layer of tissue called stratified squamous epithelium, which is very impermeable to drugs.

Injecting drugs into the esophageal tissue is another option, but that is uncomfortable for patients and inconvenient because it has to be done at a doctor’s office. There is also at least one anti-inflammatory steroid drug that is formulated as a thick mixture, allowing it to remain in the esophagus longer after being swallowed, but the drug still has some difficulty passing through the impermeable squamous layer.

In this study, the researchers set out to develop new drug formulations that would include molecules that could increase the permeability of those esophageal cells, allowing more of the drug to pass through.

To identify molecules that would enhance permeability, the researchers designed a screening system that mimics the structure of the esophagus. This system contains esophageal tissue pressed between two vertical plates. Drug formulations can be poured into the top of the system, simulating oral ingestion. The researchers can then measure how much of the drug passes through the tissue and is collected by wells in one of the plates.

Using this system, the researchers were able to measure how different excipients — inactive ingredients that help enhance drug effects — affect the permeability of the esophageal tissue. First, they tested about 100 different compounds and identified several top candidates. Then, they tested pairs of these excipients and found that the most effective combination was a pair of bile salts called sodium chenodeoxycholate and sodium cholate.

These salts appear to work together to loosen up the cell-cell junctions that normally act as a barrier to drug molecule entry. The researchers added those bile salts to a polysaccharide-derived hydrogel, which has a viscous consistency that allows it to lightly coat the lining of the esophagus.

“The hydrogel helps the formulation remain on the esophageal surface for longer, while the bile salts help increase transport across the tissue,” Karavasili says. “Our data suggest that the bile salts temporarily loosen these cell–cell junctions, mainly by interacting with calcium ions that help maintain junction integrity. This creates a more permissive pathway between the cells, allowing larger molecules to move into the mucosal tissue more efficiently.”

Minimizing side effects

In tests in animals, the researchers showed that this formulation could be used to effectively deliver infliximab to the esophagus. They also found that the loosening of the cell-cell junctions was temporary, and the cells returned to normal within three days.

This kind of delivery could help to avoid the side effects that patients sometimes experience when infliximab is given systemically, the researchers say.

“We were interested in delivering anti-TNFs as a model drug, but also to help people who suffer from conditions like Crohn’s disease to have options that could be delivered to the site,” Traverso says. “If we have the possibility of site-directed delivery, we may be able to mitigate systemic side effects from these immunosuppressing agents.”

The researchers are now working on further optimizing the formulation for potential testing in humans. One key goal is to ensure that the gel adheres for long enough to deliver the drugs, but not so long as to cause discomfort for patients. The researchers are also exploring the possibility of using this approach to deliver other types of drugs.

“This is a platform to enable the development of drug-delivery systems for the esophagus, which hasn’t been possible before because the tools haven’t existed,” Traverso says.

The research was funded by the Karl van Tassel Career Development Professorship, the Department of Mechanical Engineering at MIT, the Division of Gastroenterology at Brigham and Women’s Hospital, and the U.S. Advanced Research Projects Agency for Health (ARPA-H), which notes that the views and conclusions contained in this article are those of the authors and should not be interpreted as representing the official policies of the United States government.

When it comes to predicting people’s preferences, it pays to consider “the power of three”

Steve Nadis | MIT Laboratory for Information and Decision Systems — Thu, 11 Jun 2026 15:10:00 -0400

In his 1927 paper, “A law of comparative judgment,” the American psychologist L. L. Thurstone proposed that when people select one option among multiple alternatives, they are picking the one that has the highest value to them, even though they cannot assign a particular number to that choice.

Thurstone was a pioneer of “psychometrics” — a field built upon the premise that mental processes, which we cannot see, can nevertheless be measured and quantified. His 1927 paper laid the groundwork for what are now called random utility models, which provide a mathematical framework for describing human preferences — information that can be relied upon, in turn, to make predictions about various hypothetical situations.

Random utility models (RUMs) are so named because they assess the “utility,” or benefit, that can be obtained from a given choice — such as deciding which book to read first among the stack of novels you brought back from the library. “These models are inherently random,” explains Gabriele Farina, an assistant professor in MIT’s Department of Electrical Engineering and Computer Science (EECS) and principal investigator at the Laboratory for Information and Decision Systems (LIDS), “because people are different. Everyone has their own preferences, and even those preferences can vary from time to time.” For example, someone who normally picks coffee over tea in the morning, and prefers tea after dinner, may, upon occasion, mix up that order entirely.

RUMs, to be sure, are frequently used within government and industry in situations of far greater consequence than the selection of a hot (or iced) beverage. The models routinely facilitate predictions regarding what people will elect to do in so-called counterfactual (“what-if”) scenarios such as: How will they get to work or school if a major thoroughfare is shut down for construction? What routes and modes of transport will they take? Or, if a city suddenly receives a windfall of $20 million, how should those funds be disbursed to maximize the common good?

Given that RUMs have been with us for almost 100 years, growing in sophistication over time, one might imagine that, at this stage, there would be little room for improvement. That, however, is not the case.

A paper presented in April at the International Conference on Learning Representations in Rio de Janeiro, Brazil, uncovered basic facts that show there is much more to be gleaned from these models than had traditionally been supposed. The paper was authored by Yeshwanth Cherapanamjeri, a former MIT postdoc now based at Nanyang Technological University in Singapore; Farina, also core faculty in MIT’s Operations Research Center (ORC); Constantinos Daskalakis, the Avanessians Professor of Computer Science at MIT and a member of MIT's Computer Science and Artificial Intelligence Laboratory; and Sobhan Mohammadpour, an MIT PhD student in computer science based at LIDS and EECS.

The group’s findings stem, in part, from a deficiency in the way RUMs are commonly estimated in practice, which has persisted since the days of Thurstone. The data upon which the models are estimated have been largely drawn from so-called pairwise-comparisons: In a choice between items A and B — whether it pertains to movies on Netflix, competing products on Amazon.com, news stories posted on Google, and so forth — which one would you pick? One reason this approach has been so pervasive, explains Daskalakis, is that “assigning a precise numerical score, such as 4.37, to the benefit you get from a single item is very hard. Whereas comparing two things, and deciding which one you like better, is cognitively much easier to do.” But therein lies the rub, he adds. “With this way of assessing people’s preferences, looking at just two things at a time, it is impossible to find correlations between the numerous choices.”

The standard way of applying RUMs assumes that the utilities derived from A and B are independent, but they may, in fact, be linked, and that would be important to know. If someone campaigning for elective office finds out that a potential voter favors gun control, for instance, there is a reasonable chance that same person also favors government-sponsored child care. Similarly, a fan of independent movies might also be partial to foreign films, but less enthusiastic about Hollywood action blockbusters. “If a digital platform has a blind eye to the existence of such correlations, it will not be able to estimate preferences very accurately,” Daskalakis notes. “And if Netflix regularly shows you an assortment of movies you don’t care about, you might sign off and cancel your subscription.”

The MIT team proved that it is impossible to get information about correlations from two-way comparisons alone. Correlations can be discerned, however, when large numbers of people rate three alternatives in their order of preference. The same information can also be obtained from a combination of best-of-three and best-of-two choices. In practice, Mohammadpour explains, “you would get a bunch of people to rank three items. You could then utilize the method we developed for merging those individual results into one big model that can provide us with the big picture.”

Their research effort, according to Farina, is focused on the computational side of RUMs, devising algorithms that can extract preference information and figuring out how much data is needed to do so or, equivalently, how many experiments need to be run. The good news, he says, is that efficient algorithms are, indeed, possible for this purpose. The requisite number of experiments does not grow exponentially with the number of items in the catalog or database that’s under review.

“This paper provides a crucial breakthrough,” comments Emma Frejinger, a computer scientist at the University of Montreal. “It mathematically proves why traditional data collection fails and demonstrates that simply asking users for their best-of-three [choices] unlocks the ability to accurately train these powerful models. This finding provides a highly practical roadmap for collecting better data to drive more accurate optimizations.”

“Building utility models is going to remain a very active area,” Daskalakis insists. “Just as RUMs have been critical to the internet economy since the late 1990s, they are, and will remain to be, critical to the alignment of AI models going forward.” More importantly, he adds, “RUMs play a central role in the commercial viability and usefulness of large language models [LLMs].” During the training period, people are typically asked to rank the various candidate outputs of these LLMs, from which the models can gain a better sense as to the kind of text — in terms of tone, style, and content — that is preferred.

Given that we’re constantly “besieged with a vast sea of options in so many different domains,” Daskalakis says, “you cannot possibly ask people to communicate all their personal preferences for all possible scenarios. So what you can do instead is build a model that predicts what people think about the different possible outcomes. And you have to keep improving and updating your model in an iterative process until, hopefully, you can make good predictions.”

A shot of carbon dioxide rewires how cement sets

Andrew Paul Laurent | MIT Concrete Sustainability Hub — Thu, 11 Jun 2026 12:10:00 -0400

One September day, it started to snow inside MIT’s Pierce Laboratory.

Researchers depressurized a tank of liquid carbon dioxide (CO₂), instantly freezing it and releasing solid flakes. These were blended into cement paste and pressed into discs roughly the size of a dime, each sealed with a thin layer of vegetable oil to keep water in and air out. The team trained lasers on each, observing for the first time the transient chemical reaction that might explain why CO₂-injected cement paste gains its strength faster.

Injecting CO₂ into cement products like concrete is one way to store it and keep it out of the atmosphere. The process has attracted commercial interest, with a growing number of companies offering CO₂-injected concrete mixes. But until now, the underlying cement chemistry hadn't been directly visualized.

A new open-access paper in the Journal of the American Ceramic Society — led by Associate Professor Admir Masic and first-authored by graduate student Marcin Hajduczek, both of the MIT Concrete Sustainability Hub and MIT Department of Civil and Environmental Engineering — describes the chemical sequence that unfolds after CO₂ meets fresh cement paste. Co-authors include MIT colleagues Santiago El Awad and Franz-Josef Ulm, alongside researchers from IIT Jodhpur and CarbonCure Technologies.

Previous studies had pieced together a story about CO₂ injection’s chemical impacts from theory and indirect evidence; the key reactions simply moved too fast, and vanished too completely, for conventional techniques to catch them in the act. Raman confocal microscopy could — and it works on a simple principle: Illuminate a molecule with a laser, and the scattered light will reveal its identity. The light interacts with each material’s unique chemical bonds, shifting in energy to produce a distinct spectral “fingerprint.” Even the most fleeting and amorphous phases leave a readable trace.

“We’ve used Raman spectroscopy to better understand some of the most interesting materials in history, from the Dead Sea Scrolls to Ancient Roman concrete,” says Masic. “Cement paste may seem less glamorous in comparison, but pointing a laser at CO₂-injected cement paste as it hardens allows us to visualize things that haven’t been seen before.”

What they saw, unfolding during 24 hours of continuous scanning, was a three-act chemical drama.

Act One: Capturing calcium

The moment that CO₂ is added to the fresh cement paste, it goes to work. It dissolves into the pore solution and reacts with calcium released by the dissolving clinker, precipitating as various forms of calcium carbonate. Clinker is produced by heating limestone and aluminosilicate materials in a kiln, forming the primary ingredient ground into a fine powder to make cement. This happens within the first hour, temporarily slowing the normal hydration reaction, which requires calcium to proceed.

In contrast, when CO₂ is not present, the calcium released by the dissolving clinker remains available locally, supporting the gradual formation of the material’s binding phases as it sets.

Left without calcium, the silicates released by the clinker dissolve into the pore solution and precipitate far from their source, linking together into chains that form an interconnected silica gel network throughout the paste. This amorphous, fleeting gel sets the stage for what follows.

Act Two: The ghostly gel

Once the injected CO₂ is fully mineralized — around four to five hours after mixing — normal hydration resumes. Calcium hydroxide begins to precipitate into the pore space, and when it does, it encounters the silica gel network waiting for it.

The reaction between the two phases begins immediately, producing calcium silicate hydrate (C-S-H), the compound that gives cement its binding ability. What makes this form of C-S-H distinct is where and how it forms: not clustered around clinker particles as in conventional hydration, but distributed throughout the entire matrix, wherever the silica gel had spread.

The CO₂ had temporarily suppressed the paste’s alkalinity, and that lower pH was the only thing keeping the silica-gel intact. As hydration reasserts itself and produces standard hydration products, namely C-S-H and calcium hydroxide, the latter drives pH back up to typical levels in a self-reinforcing loop; the silica-gel reacts with calcium hydroxide through a so-called pozzolanic reaction. Within eight hours, the silica gel is almost entirely gone — the previously well-distributed gel network turns rapidly into additional C-S-H during this critical early window.

“At first, the fleeting nature of the silica gel looked like a fluke in the Raman data. But it quickly became clear that its sudden disappearance was a consistent, undeniable feature of every CO₂-injected sample,” says Hajduczek.

Act Three: A rewired matrix

With the silica gel consumed, the paste settles into conventional hydration, but what it leaves behind is measurably different. Because the new binder was distributed more evenly throughout the cement matrix, the resulting microstructure is stronger and more uniform at an early age. In the study, paste mixed with CO₂ at 1 percent by cement weight achieved, on average, 13 percent higher compressive strength at 24 hours, compared to reference mixes.

“We’ve been injecting CO₂ into cement products for years without fully understanding what it was doing inside. Now that we can see it and understand the underlying mechanism that leads to improved performance, we can start to control it. And there’s a lot of room to push,” says Masic.

The findings also refine a leading explanation for CO₂-injected cement paste’s higher early age strength: the calcium carbonate crystals, previously suspected to seed C-S-H growth, turn out to be passive bystanders embedded in the silica gel template rather than reacting to form C-S-H.

Where the chemistry goes next

Knowing the mechanism gives researchers a more specific set of questions to pursue. The silica gel template explains the distribution of the new C-S-H, but directly measuring its mechanical properties remains a next step.

On the practical side, dosage matters: Flood the system with too much CO₂ and calcium gets locked into carbonate before the gel can form and react. If the paste used here forms abundant C-S-H, it could theoretically offset up to 40 percent of the carbon emissions from cement production, excluding emissions associated with the fossil fuels used in the process. In practice, however, the achievable offset is likely to be only a fraction of that value, although still potentially significant.

But even with these open questions, the ghostly gel has been caught. And now that researchers know what to look for, the chemistry that unfolds in those first eight hours is no longer invisible.

Would you return a favor? Scientists say it depends on the relationship

Anne Trafton | MIT News — Thu, 11 Jun 2026 00:00:00 -0400

When a friend buys you a cup of coffee, it’s likely that next time, you’ll return the gesture. This type of reciprocal generosity has been well-documented in behavioral economic studies.

However, anthropologists and other social scientists have known for decades that in the context of relationships where one person has more power, status, or influence, reciprocal generosity is usually not the norm.

Researchers at MIT have now experimentally demonstrated, for the first time, that small changes to the relationship context can dramatically change people’s actions and expectations of reciprocal generosity.

During interactions between people of different social status, people tend to expect that generosity will flow one way, and it can be either up or down. It may be that a professor always buys coffee for her students, or that a student always offers to help carry groceries for his resident advisor. Once the precedent is established, it is expected to continue.

One interpretation of the findings is that keeping track of whose turn it is to do a favor is the exception in social interactions, not the rule. That is, it is extra work that we do when we want to maintain equal relationships.

“In many intimate relationships, hierarchical relationships, or other kinds of role-based relationships, you don’t put in the work of trying to keep track of turns,” says Rebecca Saxe, the John W. Jarve Professor of Brain and Cognitive Sciences, a member of the McGovern Institute for Brain Research, and associate dean of science at MIT. “Under this interpretation, we just follow precedent because following a precedent is easier. We all know what to expect, and we don’t have to keep track of what happened last time.”

Saxe is the senior author of the study, which appears in the journal Open Mind. MIT graduate student Alicia Chen is the paper’s lead author.

Changing expectations

Most experimental studies of generosity have been done in the context of behavioral economics and game theory. In such experiments, people are usually paired with a stranger and asked to play games that require coordination. Such studies have found that people tend to use turn-taking and reciprocity as their default strategies. These scenarios, however, are stripped from any social context that might exist between people in the real world.

Saxe and Chen wanted to see if they could measure the effects of social context by incorporating relationships into the type of experiments used to evaluate people’s expectations regarding generosity.

“Where generosity becomes hard and complicated is when it starts to occur in the context of existing relationships, because it changes the terms of the relationships,” Saxe says. “What’s expected of you is very different within a relationship than outside of one.”

To study these effects, the researchers designed experiments in which participants read stories about different types of interactions. In some of the scenarios, the subjects of the stories were described as having either symmetric or asymmetric relationships. In others, they were given specific social relationships such as aunt-niece or manager-employee.

Each story described interactions that might be seen in typical daily life, such as buying coffee for a co-worker or preparing a meal for one’s family. Participants were then asked to predict what would happen the next time the interaction occurred.

In all of these scenarios, the researchers found that people expected that generous acts would be reciprocated when they occurred between individuals in symmetric relationships such as friends, cousins, or co-workers of equal rank. However, their expectations changed for asymmetric relationships, where each person has a different social status. In those cases, people expected that any precedent that was set would continue in the future.

One possible explanation for this is that reciprocity is not the norm but an exception that only occurs in the interactions between equals or strangers, the researchers say. Many of our interactions are with people with whom we have asymmetric relationship, and to maintain those relationships, it’s simply easier to follow precedent.

“If there’s no need to keep track of our equal status, then in some ways it’s the default to fall back on following precedents,” Saxe says.

Maintaining relationships

The study showed that in asymmetric relationships, generosity could flow in either direction. Once that direction was established, it was expected to continue. For example, after an older brother bought concert tickets for a much younger brother, the study participants expected that the older brother would also buy the tickets for the next concert.

“We found that when people know the relationship is asymmetric, they don’t expect reciprocity; they expect the same action to keep on going,” Chen says. “If the lower-rank person acts generously, people expect that to continue, and if the higher-rank person acts generously, people expect that to continue.”

Following precedents is not only easier, but keeping up these actions may help solidify and define existing relationships. For example, anthropologists have long known that gift-giving helps to construct and maintain social relationships.

“Following a precedent can be a way of actively maintaining relationships and hierarchies, when the asymmetry of the exchange truly reflects the asymmetry of the relationship,” Saxe says.

The researchers are now working on creating computational models that could be used to analyze different factors that people take into account when they’re considering whether someone might reciprocate a generous act. In addition to the factors examined in this study, others could include how much each person will benefit, what type of relationship they’re in, and culturally specific expectations of how people should act in different situations.

“One really powerful thing about these models is that we can build in existing theories, add things to the models, and then compare how much these extra factors, like considerations related to social relationships, matter in terms of explaining what people are doing,” Chen says. “This allows us to quantitatively compare the different theories to each other.”

The research was funded by the Simons Foundation Autism Research Initiative and the Patrick J. McGovern Foundation.

New imaging system sees through murky waters

Jennifer Chu | MIT News — Thu, 11 Jun 2026 00:00:00 -0400

For remotely operated underwater vehicles, cloudy and turbulent waters are often a no-go. When vehicles settle on the seafloor or dig through a sandbed, they can kick up clouds of sediment that make it tough for onboard cameras to see through. Often, the only thing to do is to wait until the marine dust settles before a vehicle can safely proceed.

But a new underwater mapping technique developed by engineers at MIT and the Woods Hole Oceanographic Institution (WHOI) may allow vehicles to see through murky, low-visibility waters.

The method fuses visual images from optical cameras with acoustic data from sonar sensors. The combination enables a vehicle to quickly map the general shape of its surroundings using sonar, even in low-visibility waters. A vehicle can move toward certain shapes in the sonar-mapped environment, coming close enough for optical cameras to visually resolve specific objects in detail.

The technique is akin to pairing a dolphin’s echolocation with a sea turtle’s close-range vision to see and navigate through murky water, in real-time.

The researchers tested the method in tank experiments where they could control the water’s degree of visibility. Even in the cloudiest conditions, the system was able to see through the sediment to map the tank’s environment and visualize centimeter-scale details of objects in the tank.

The team is further improving the technique, which they’ve named Sonar-MASt3R. They envision that the mapping method could safely guide underwater vehicles through murky environments for a range of applications, including scientific exploration, underwater construction and maintenance, and deep-sea recovery.

“We hope that this work enables us to do more operations in those challenging, low-visibility environments, and helps provide more coverage in areas that are difficult to operate in today,” says Amy Phung, a graduate student in MIT’s Department of Aeronautics and Astronautics, who led the work.

Phung presented a paper detailing Sonar-MASt3R this week at the IEEE International Conference on Robotics and Automation (ICRA). The paper’s co-author is Richard Camilli, senior scientist of applied ocean physics and engineering at WHOI.

The best of both

To see underwater, scientists have generally taken an either/or approach, using either optical cameras or sonar sensors to guide the way. Optical cameras can provide detailed visual imagery of a scene, but only in waters that are relatively clear and well-lit. In contrast, sonar sensors perform just as well in clear and murky water; by emitting acoustic waves and measuring the time and angle at which they return, sonar sensors can determine the exact shape, distance, and depth of objects in the environment, though a sonar map lacks any visual detail.

To get the best of both modes, scientists have looked to combine the two in a new approach known as “opti-acoustic fusion.” In a handful of prior works, research groups have merged sonar and optical data in mapping techniques that are mostly geared toward object recognition and reconstructing workplace environments. Most techniques require time to sync and process the data and therefore do not work in real-time, while only a few can map an environment in 3D. None have been applied to high-resolution mapping underwater in murky, turbid conditions.

Phung, who is a student in the MIT-WHOI Joint Program, and Camilli, her advisor, aimed to develop an opti-acoustic fusion technique that would generate detailed 3D maps of underwater environments in real time and in low-visibility conditions. The team was motivated, in part, by challenges in safely recovering unexploded underwater mines.

“There can be old explosives in areas that make it unsafe for ships to be in, and the ability to get rid of those safely is best done by robotics,” Camilli says. “But a lot of these explosives are set in surf zone environments where visibility adds to the challenge of doing this safely. That’s one of many applications that our technique can be used for.”

Cloudy, with a chance of mapping

The new method, Sonar-MASt3R, builds on an existing technique, MASt3R, that was developed by researchers in France. MASt3R is an image matching algorithm that is trained to take in visual images of the same scene and quickly estimate the relative depth of each pixel in the scene. In this way, MASt3R can generate a 3D map of the environment in real-time, based on a camera’s 2D images.

“The downside is that there is no sense of scale,” Phung says. “It will say ‘this pixel is five units closer than this pixel,’ but it can’t say whether that’s 5 meters or 5 feet.”

Luckily, sonar provides absolute measurements of scale. The timing of sonar reflections can be translated directly into a specific depth and distance of objects that the signals bounced off, as well as their shape and contour.

In their new work, Phung and Camilli used sonar data to correct MASt3R’s scaling and generate precise 3D maps of underwater environments. Even in murky water, the method’s sonar-corrected map would enable a vehicle to know the precise location of objects, and therefore how far to safely move in for a closer inspection, which the vehicle could then do using conventional optical cameras.

The team tested Sonar-MASt3R in experiments with a tank that they filled with water, sediment, and a variety of objects such as a small boulder, a coffee mug, and a packing crate. Inside the tank, they also set up a robotic arm, onto which they mounted an underwater camera, and a sonar sensor.

For each experimental run, they first carried out a sweep trajectory, in which the robotic arm slowly swept from one side of the tank to the other to capture sonar and visual data. With this first sweep, Sonar-MASt3R quickly creates a coarse sonar-based map of the shapes and contours of the tank and its objects. The coarse map is then used to record close-up camera images of the objects, which are used to improve the map resolution. A “keyframe” approach quickly compares each new image frame to the last keyframe. If a frame provides new information not contained in the last keyframe, the image is added as a new keyframe to the map. If it is similar, it is immediately discarded. In this way, the approach can quickly fill in the map with relevant visual detail, in real-time.

The researchers tested their new approach underwater, testing eight different levels of turbidity, which they created by stirring up the tank’s sediment. Compared with other opti-acoustic fusion approaches, Sonar-MASt3R generated more accurate 3D maps and resolved smaller, centimeter-scale details, and in cloudier conditions. In the cloudiest condition, which the robotic arm’s cameras could not see through, its sonar sensors were able to generate a rough map of the tank’s hidden objects. This initial map enabled the arm to move safely through the murk and closer to specific objects, which its underwater camera could then visualize in more detail.

“An analogy would be if you were to go into a china shop in the dark, and try to pick your way around to find a specific coffee mug without knocking things over,” Camilli offers. “This would allow you to do that.”

The team plans to test the approach in natural underwater conditions, where they suspect that the mapping task should be more straightforward.

“In a tank, it’s like an echo chamber,” Camilli says. “It’s like trying to do this in a funhouse mirror setting where you get all these distortions and reverberations and ghost images that really complicates the processing. If you put it in the real world, it should be easier.”

Then, they say, Sonar-MASt3R could help scientists safely explore in cloudy, turbid, and murky underwater regions.

“The real value in this effort is so we can use this technology in mission scenarios that are untractable right now,” Phung says. “And there are plenty of untractable missions because we don’t have the observational or perception capabilities.”

This research was supported, in part, by NASA, and the National Science Foundation.

Myriam Heiman named director of The Picower Institute for Learning and Memory

School of Science — Wed, 10 Jun 2026 15:25:00 -0400

Myriam Heiman, the John and Dorothy Wilson Professor of Neuroscience at MIT, will become the director of MIT’s Picower Institute for Learning and Memory, effective July 1. She succeeds Picower Professor Li-Huei Tsai, who is stepping down after leading the institute for 16 years.

Heiman, a molecular neurobiologist and geneticist, studies the neurodegenerative diseases of the brain’s basal ganglia, including Huntington’s disease and Parkinson’s disease. Using cutting-edge techniques, including single-cell genomics and a powerful transcriptomic technique she helped invent, called translating ribosome affinity purification, she aims to understand the molecular changes that eventually lead to cell death in these diseases.

“Myriam is an extraordinary scientist, a proven leader within MIT, and a deeply caring and generous mentor. Her research to determine why specific brain cell types are particularly vulnerable to diseases such as Huntington’s has produced studies that are both deep in their insight and sweeping in their scope,” says Nergis Mavalvala, dean of the MIT School of Science and the Curtis and Kathleen Marble Professor of Astrophysics. “I firmly believe that Myriam will be an excellent leader during the Picower Institute’s next chapter.”

“I am honored to take on this role to support the institute’s exceptional scientists and trainees as they pursue discoveries that deepen our understanding of the brain and improve human health,” says Heiman, a professor in MIT’s Department of Brain and Cognitive Sciences (BCS).

The Picower Institute is a community of 16 neuroscience labs dedicated to understanding the mechanisms that drive learning and memory and related functions such as cognition, emotion, perception, behavior, and consciousness. Institute neuroscientists explore the brain and nervous system at multiple scales, from genes and molecules to cells and synapses to circuits and systems, producing novel insights into how disruptions in these mechanisms can lead to developmental, psychiatric, or neurodegenerative disease.

Picower Professor Susumu Tonegawa founded the institute as a center in 1994 before a transformative gift from Barbara and Jeffry Picower enabled it to become an institute in 2002. Li-Huei Tsai has served as director since 2009, but announced in March that she would step down after more than 16 years to focus on her research.

Heiman joined the Picower Institute, BCS, and the Broad Institute of Harvard and MIT in 2011, after completing her postdoctoral training at The Rockefeller University. She holds a PhD from Johns Hopkins University and a BA from Princeton University.

“Ever since joining the institute, Heiman’s research has been guided by the principle that fundamental understanding can lead to breakthroughs in addressing disease,” Tsai says. “Myriam has made it her mission to address these kinds of urgent questions in neuroscience.”

Heiman employs sophisticated DNA and RNA analysis technologies to gain detailed insights into how brain cell states change amid disease, revealing molecular pathways that contribute to the particular vulnerability of different cell types. In 2020, Heiman published the results of an innovative in vivo screening of every mouse gene’s impact on the survival of neurons in the brain, identifying hundreds necessary for sustaining neurons and highlighting a specific gene that promoted their resilience in the context of Huntington’s disease.

Other studies, both in mice and in postmortem human brain samples, have revealed errant immune responses in neurons and in the brain’s blood vessels that contribute to the disease’s progression. The latter finding arose in a 2022 paper, published with MIT Computer Science and Artificial Intelligence Laboratory colleague Manolis Kellis, that also provided the field one of the first cellular atlases of the brain’s vasculature.

Her research has also produced insights into other neurodegenerative and psychiatric disorders, including ALS and frontotemporal dementia. In 2024, together with Kellis, Heiman published a paper in Cell showing the diseases have remarkable overlaps at the cellular and molecular levels, revealing potential targets that could yield therapies applicable to both disorders. Heiman’s latest research is also producing new insights into substance use disorders and schizophrenia.

Her research program has garnered many awards. In 2021, Heiman became co-recipient of a National Institutes of Health Transformative Research Award, which “promotes cross-cutting, interdisciplinary approaches that could potentially create or challenge existing paradigms” as part of the NIH’s High-Risk, High-Reward Research program. The next year she also received a prestigious NIH R35 grant to find early triggers of disease progression.

Heiman is also a dedicated teacher and mentor. In 2017, she earned the Department of BCS award for excellence in graduate mentoring; and in 2020, she received the department’s award for excellence in undergraduate teaching. In 2024, she was named one of 23 faculty across MIT who are “committed to caring” — an award given out by MIT’s Office of Graduate Education to faculty members who have served as exceptional mentors to graduate students.

Beyond MIT, Heiman serves on editorial boards and the scientific advisory board of the nonprofit Huntington’s Disease Foundation, an organization that supports research aimed at finding treatments and a cure for Huntington’s and related disorders..

Heiman says she is looking forward to her new role in service to MIT by leading the Picower Institute.

“I approach this role with humility and enormous enthusiasm,” Heiman says. “The Picower Institute has an extraordinary legacy, and I’m eager to do everything I can to help support the next generation of transformative research.”

To study how chips really work, MIT researchers built their own operating system

Rachel Gordon | MIT CSAIL — Wed, 10 Jun 2026 13:35:00 -0400

A new kernel, or core program within an operating system, gives researchers a cleaner view of what’s happening inside a processor. Called Fractal and developed at MIT, the kernel has already surfaced previously unknown behavior in Apple’s M1.

When security researchers want to understand what a modern processor is really doing with the kind of detail that determines whether attacks like Spectre and Meltdown are possible, they usually run their experiments on top of an operating system that was never built for the job. They open up macOS or Linux, patch the kernel by hand, and hope the modifications hold. The approach is unstable, hard to reproduce, and on Apple’s platforms, slated for deprecation.

A team at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) decided to build something different. Fractal, an operating system kernel written from the ground up, treats the hardware itself as the object of study. Its first major use, a deep look at branch predictors — a CPU’s way of guessing what code to run next, before it knows for certain, so it doesn’t have to waste time waiting to find out — inside Apple’s M1 processor, has already turned up findings that prior work missed, including the first evidence that a class of speculative attack known as “Phantom” affects Apple Silicon.

“We’re using hardware in ways it wasn’t designed for,” says Joseph Ravichandran, the MIT PhD student in electrical engineering and computer science (EECS) who led the project. “It’s not even obvious that this is a possible thing you could do with the hardware. But we found a way to pull all these different primitives off. It’s like a microscope. If you’ve got a hand magnifying glass, you can see a little bit. But if you had an electron microscope, now we’re really talking. That’s what Fractal is. The electron microscope of operating systems.”

A clean room for chip research

The core problem Fractal solves is one that researchers have worked around for years. Modern processors keep state in many internal structures: branch predictors, caches, translation lookaside buffers, and more. To study how those structures behave across the boundary between user code and kernel code, two domains the chip is supposed to keep isolated, researchers need to run nearly identical experiments on each side of that boundary. On a general-purpose operating system, that is very difficult. The system itself manages privilege levels, address spaces, and scheduling, and it injects its own activity into every measurement.

Fractal inverts the model. It boots directly on bare metal, with no other software running, and exposes primitives that let a single experiment switch privilege levels at runtime while executing the same instructions in the same address space. The team calls the underlying technique multi-privilege concurrency, and it relies on a new construct they introduced: the outer kernel thread, which sits inside a user process’s memory but executes with kernel privileges.

The result is an experimental setup with almost no background noise. Where measurements taken under macOS or Linux are blurred by interrupts, scheduler activity, and address-space management, Fractal produces flat baselines and clean signals.

What Fractal found on the M1

Apple’s M1 implements an ARM specification called CSV2, which is supposed to prevent code running in one privilege level from steering speculation in another. Using Fractal, the MIT team confirmed that the protection works for the execute stage of indirect branch prediction: a user-mode program cannot make the kernel speculatively execute a chosen target through the indirect branch predictor.

But the team also found something the chip’s designers may not have intended. The CPU still fetches the target into the instruction cache before the protection kicks in. That fetch is observable through a side channel, which means user code can still influence what the kernel pulls into its caches across the privilege boundary. The same pattern appeared between processes assigned different address space identifiers.

The team also produced the first evidence that Apple Silicon exhibits Phantom speculation, a class of misprediction previously demonstrated only on AMD and Intel processors. In Phantom, ordinary instructions, including a no-op, can be misinterpreted by the CPU as branches, triggering speculative behavior the program never asked for. On the M1, Fractal showed that Phantom fetches succeed across both privilege levels and address spaces, though the execute phase remains blocked.

A separate Fractal experiment overturned a finding from earlier work on the M1’s conditional branch predictor, which had reported that cross-privilege training worked on Apple’s performance cores, but not its efficiency cores. The Fractal team showed that the conditional branch predictor has no privilege isolation at all, on either core type, and that the earlier result was likely an artifact of macOS quietly migrating threads between cores during system calls.

“For us, it is a true independent variable,” Ravichandran says. “You change the privilege level, nothing else changes. The only thing that could explain whether the attack succeeds or not is the privilege level.”

A tool, not a one-off

Fractal supports x86_64, ARM64, and RISC-V, and consists of more than 31,000 lines of code. The team designed it as infrastructure rather than as a single experiment, with familiar POSIX system calls, a C library, and ports of standard tools like vim, GCC, and the dash shell, so that researchers can move existing experiment code over with minimal friction.

The MIT team disclosed its M1 findings to Apple’s product security team. In an unusual reversal, Apple’s engineers also examined Fractal.

The longer-term ambition is bigger than any single result. Ravichandran wants Fractal to become to microarchitecture research what tools like QEMU and FFmpeg are to their fields: shared infrastructure that the whole community builds on.

“My hope is that our results as a community get significantly more reliable, significantly more accurate,” says Ravichadran. “With this reduced noise, this clarity, and this guarantee that you’re running on the right core, on the right system.”

“Fractal is a strong architecture contribution because it turns an often ad hoc microarchitectural reverse-engineering workflow into reusable research infrastructure,” says University of Southern California assistant professor Mengyuan Li, who wasn’t involved in the paper. “By reducing software noise and giving researchers tighter control across privilege boundaries, it makes difficult hardware experiments much easier to interpret.”

Ravichandran worked with Mengjia Yan, an MIT associate professor of EECS and CSAIL principal investigator, on the paper. Their work was supported, in part, by the National Science Foundation, the U.S. Air Force Office of Scientific Research, and ACE, which is part of a program sponsored by the U.S. Defense Advanced Research Projects Agency. They presented their work at the IEEE Symposium on Security and Privacy in San Francisco, California.

Augmented reality system could make medical ultrasounds easier to interpret

Anne Trafton | MIT News — Wed, 10 Jun 2026 05:00:00 -0400

Interpreting medical ultrasound images is a difficult task, requiring a technician to look at 2D images and mentally arrange them into a 3D representation of what the tissue looks like.

To make that job easier, MIT researchers developed a new approach to ultrasound imaging that allows the user to visualize a 3D augmented-reality image of the object being scanned. Using a virtual-reality headset, they can see a precise 3D digital representation of what the object actually looks like, making it easier to identify and analyze.

This technique could help speed up the training process for ultrasound technicians and other health care providers who use ultrasound. It could also be deployed for use in hospitals, for tasks such as using ultrasound to place a needle in the right location for a biopsy.

“For training, this could make ultrasound more intuitive and more understandable. On the clinical side, it could be less time-consuming, more accurate, and also give health care providers more peace of mind. They wouldn’t have to wonder if they missed anything,” says Canan Dagdeviren, an associate professor of media arts and sciences at MIT and the senior author of the study.

MIT graduate students Jason Hou and Shrihari Viswanath are the lead authors of the paper, which appears today in Nature Communications Engineering. Other authors of the paper include Bowen Wu ’24 and two MIT Summer Research Program students, Cinay Dilibal, a senior at Dartmouth College, and Tanisha Shende, a senior at Oberlin College.

3D representations

Ultrasound imaging works by bouncing high-frequency sound waves off tissues in the body, which are then reflected back to an ultrasound transducer. The transducer converts these sound waves to electrical signals, which are used to create a 2D image of the tissue. Ultrasound technicians are trained to convert these images into a 3D mental representation of the tissue.

“It's a difficult skill to master, and there are long learning curves,” says Hou. “The hardest thing is this mental tomography bottleneck where you’re trained to reconstruct the 2D slices in your 3D mental space. That is a cognitive burden that can lead to inaccuracies in scanning.”

To reduce that cognitive load, the MIT team thought it could be helpful to combine two technologies: 3D ultrasound imaging and augmented reality (AR).

Three-dimensional ultrasound imaging is occasionally used in fields such as fetal imaging and echocardiography, which is used to image the heart, but most 3D ultrasound imaging systems are expensive and not widely available. For this study, the MIT team used a real-time 3D system they developed recently for use in breast-cancer detection.

Their new system includes an ultrasound probe, slightly smaller than a deck of cards, that transmits information using a chirped data acquisition system (cDAQ). The probe contains an ultrasound array arranged in the shape of an empty square, a configuration that allows the array to take 3D images of the tissue below.

Because this system has fewer ultrasound elements than a typical 3D ultrasound system, it requires less power and is less expensive to build.

The data collected by the ultrasound probe can then be compressed and streamed into a 3D computer graphics engine called Unreal Engine, which converts the voxel data from the ultrasound image into a direct 3D representation of the object, with no loss of information. Wearing an AR/VR headset, the user can see this 3D rendering representing the internal structure, superimposed over the object’s actual location — like X-ray vision. By tilting their head or approaching from a different direction, the user can see different views of the object, making it easier to identify.

Easier to use

The researchers tested their new technology, which they call AR-VIU (augmented real-time volumetric imaging in ultrasound), with a group of 18 participants. Nine of the subjects were experts in ultrasound technology (including sonographers and physicians), and nine had never used ultrasound before.

Each user performed identification tasks using four different ultrasound technologies. In one condition, they viewed 2D images on a regular screen, which is the way that most ultrasounds are now performed. They also viewed 3D images on a regular screen, as well as two augmented reality conditions: one 2D and one 3D (AR-VIU).

In one round of experiments, users were asked to identify an object embedded in gelatin — such as a spring, a ball, or a screw — inside an opaque container that was scanned with ultrasound. In a second set, they were asked to use a pen to mark the location of “tissue phantom” — a gel-like material engineered to mimic human tissue. This simulates the task of locating the right spot for a needle during a biopsy.

The researchers found that the AR-VIU system significantly improved all users’ ability to identify and locate objects. The effect was especially strong for novices, who performed nearly as well as experts when using AR-VIU. When using the traditional 2D imaging system, experts performed much better than novices.

“Overlaying images with the anatomy and providing 3D visual context makes ultrasound significantly easier for novices to understand,” Viswanath says.

In interviews after the experiments, most of the novices reported that they preferred the AR-VIU approach, with many saying that it made the tasks easier.

“The 3D system imposes less brain drain, it’s more intuitive, and it’s easier to understand what is happening in the targeted region,” Dagdeviren says.

Many of the experts said they preferred the traditional 2D imaging because that is what they were accustomed to and had been trained to use. However, those experts also said they could see the benefits of the AR-VIU system in some situations, such as placing a needle for a biopsy or visualizing the movement of the heart wall during echocardiography.

The researchers are now working on further improving the resolution of the imaging and doing additional tests to demonstrate the accuracy of the AR-VIU technology.

The research was funded by the MIT Media Lab Consortium, the National Science Foundation, an MIT HEALS graduate fellowship, and an MIT-Tata graduate fellowship.

The consequences of relying on AI for accurate news

Adam Conner-Simons | MIT Media Lab — Tue, 09 Jun 2026 16:30:00 -0400

It’s no secret that the last few years have seen a massive explosion in the use of artificial intelligence for general information-gathering. An even more recent trend, though, is how large language models (LLMs) like ChatGPT, Claude, and Gemini are increasingly being used for verifying and consuming news; reports from the Pew Research Center over the last year found that one-in-five U.S. teens regularly use LLMs to get their news, while one-in-four young adults have reported using them for that purpose at least once.

A new open-access study from the MIT Media Lab should give some of those users pause: Researchers found that, over the course of a month, participants who relied on AI systems to verify facts actually got worse at detecting misinformation on their own when their chatbots were taken away.

This phenomenon, which is often referred to as the “AI dependency paradox,” has been observed in a wide range of knowledge domains, like the 2025 study that found that doctors who used AI got worse at detecting cancer on their own. The dynamic mirrors broader tech trends around so-called “deskilling” (or “cognitive offloading”) that have been well-documented for decades, from calculators weakening our math skills to Global Positioning System (GPS) technologies impacting our natural sense of direction.

In the new Media Lab study, which tracked 67 people over four weeks as they evaluated news headline-image pairs, participants were 21 percent more accurate in detecting fake news when assisted by an AI chatbot during a session — confirming previous research out of the MIT Sloan School of Management demonstrating that AI can be an effective tool in reducing people’s beliefs in false information.

However, the study showed that a new wrinkle emerged when the AI was no longer present: By week four, participants’ unassisted performance on new news items declined by 15 percentage points compared to before the study started. (Roughly a quarter of all participants actually reported feeling that they were getting better at detection, even as their performance declined.)

Dunning-Kruger creeps in

“Users get excited about these ‘magical’ LLMs, but forget that they’re just statistical models that predict the next ‘token’ in a sequence [of letters/words],” says MIT media arts and sciences (MAS) PhD student Anku Rani, co-lead author of a new paper about the research, alongside fellow MAS PhD student Valdemar Danry. “Many impressive behaviors emerge from scaling this, but it comes with real limitations, both in what the model can reliably generate and in its broader impact on the people using it.”

Qualitative analysis identified distinct behavioral patterns, with the team labeling one-fifth of all participants as "Dependency Developers” who gradually shifted from active self-reliance to passive acceptance of AI guidance.

In the post-experiment survey, one respondent explicitly acknowledged this transition, noting their passive role in the process. “While [the chatbots] did emphasize that you must check across multiple sources to make sure a story is true, they didn’t teach me much about exploring the context of the images themselves,” the participant said.

The research team said that these AI models are particularly vulnerable to mistakes in the midst of emotionally charged breaking news, as exhibited by the widespread misinformation that accompanied President Trump’s recent assassination attempt and major events during the Iranian war. (The authors also point out that the original human-created news content that’s used to train the AI models is increasingly unreliable and/or biased, further exacerbating the problem.)

The paper, which Danry and Rani presented at the 2026 CHI Conference on Human Factors in Computing Systems, was co-authored by Assistant Professor Paul Pu Liang, Senior Research Scientist Andrew Lippman, and senior author Pattie Maes, the Germeshausen Professor of Media Arts and Sciences.

The solution: Being a coach, not a crutch

The researchers say that the results of their project suggest that the specific way in which an AI interacts with a user determines whether its impact will be “as a coach, versus as a crutch.” The study found a clear distinction between conversational strategies that simply help in the moment and those that actually support active learning and skill development.

For the latter, the Media Lab team uncovered several strategies associated with stronger independent detection later on, even if the strategies initially slowed down performance during the interaction. This included the Socratic method of the AI asking guided questions, as well as so-called “deep probing,” where the system provides gently persuasive statements if the user appears to be veering away from the correct response.

“AIs that ‘tell’ by providing direct answers are more likely to foster reliance, while those that ‘ask’ via Socratic questioning are better at engaging someone to actually learn how to discern the truth on their own,” says Danry. “But it’s very much a trade-off between speed and effort.”

Rani noted a few key limitations to the one-month study, from the small dataset of roughly 50 validated news items to the demographic focus on the United States and the United Kingdom. In the future, she says that the team hopes to do similar experiments with more geographically diverse cohorts, including low-resource communities, and is also eager to explore whether other multi-modal interaction strategies — like interacting with culturally adaptive digital twins instead of text-based chatbots — help people improve their abilities to detect misinformation.

At a higher level, the researchers hope that the project will be something that educators can examine as they develop teaching plans that incorporate AI tools into their school curricula.

“It’s especially important to raise awareness in our schools and academic communities about the shortcomings of using AI as learning tools,” says Maes. “People need to know that if they ‘delegate’ their thinking, they’re not going to get better at that particular brand of problem-solving. Ultimately, the ability to question and analyze information is important for everyone, because it empowers us to solve problems and form our own independent opinions about the world.”

Danry adds that the rapidly-evolving field of machine learning and deep learning will require continuous education on the benefits and drawbacks of LLMs.

“There’s a lot of work to do in making sure that we don’t just fully offload critical tasks that we want to be able to keep on doing to these models,” he says. “We need to develop a new kind of AI literacy.”

The research project was supported, in part, by the Media Lab Consortium, an MIT Tata Center Technology and Design Fellowship, and a Google PhD Fellowship in Human–Computer Interaction.

3D-printed devices could streamline the production of drug-delivery microparticles

Adam Zewe | MIT News — Tue, 09 Jun 2026 00:00:00 -0400

MIT researchers have demonstrated a low-cost design of specialized electronic nozzles, called triaxial electrospray emitters, that could be used to manufacture time-release drug-delivery particles or self-healing materials efficiently and at scale.

Triaxial electrospray emitters use electricity to precisely dispense three liquids from microscopic nozzles to generate a steady stream with three distinct fluid layers. The liquid forms multilayered droplets, which can solidify into layered microparticles.

For instance, an array of triaxial electrospray emitters can be used to make three-layer drug-delivery nanoparticles. The outer layer might slowly erode in the stomach, revealing a second material that controls the release of a core material, which delivers medicine to a specific area of the intestines.

Developing a tiny array of electrospray emitters typically requires expensive and time-consuming microfabrication processes inside semiconductor cleanrooms, which limits their use. To overcome these drawbacks, the MIT researchers 3D-printed arrays of triaxial electrospray emitters that have 16 nozzles in an area of about one square centimeter. Each device contains an intricate network of three-dimensional microchannels that uniformly supply liquid to the nozzles.

Their one-step fabrication process takes only a few hours to produce complex emitter arrays.

When tested, the 3D-printed arrays generated uniform, three-layered droplets at scale. Such uniformity is key for high-throughput manufacturing of layered microparticles for applications like biosensors that detect chemical substances or artificial cells to aid in tissue regeneration.

“We couldn’t make a device like this in a semiconductor cleanroom. This is only possible because they are 3D-printed,” says Luis Fernando Velásquez-García, a principal research scientist in MIT’s Microsystems Technology Laboratories (MTL) and senior author of a paper describing this advance. “The particles these devices generate, whether they are used for a self-healing composite or to deliver medicine, can have a big impact in many applications. We want to democratize this technology so the benefits can touch many more people.”

Velásquez-García is joined on the paper by lead author Bryan Ivan Quintanar-Abarca of the Technological Institute of Monterrey in Mexico. The research appears in Virtual and Physical Prototyping.

A precise process

Electrospray emitters apply a high voltage to a liquid as it exits the device’s nozzle, producing a steady stream of extremely tiny droplets.

Triaxial devices contain arrays of three concentric nozzles that emit three immiscible, or non-mixable, liquids simultaneously into layered droplets, which can be used to generate compound microparticles with distinct layers.

For instance, one could use a triaxial electrospray emitter to create a biosensing particle that contains three different chemical markers, one in each layer. Electrospray emitters can make smaller microdroplets much faster than other techniques.

Miniaturization is key for electrospray devices, since the smaller the emitter, the lower the voltage required to generate droplets. The output of a single electrospray emitter is modest, so arrays of emitters are required to boost droplet production without sacrificing uniformity.

Multi-emitter electrospray devices are typically manufactured in semiconductor cleanrooms, but traditional processes limit the shapes and sizes of device components. The researchers could not find any previous reports of a miniaturized triaxial electrospray array in the open literature, highlighting the novelty of this work.

“When you build a triaxial array, you need to find a way to create geometries that have many integrated parts and extremely fine structures in the smallest footprint possible. And you need to ensure the devices will work uniformly,” Velásquez-García explains.

To do this, he and his collaborators used a 3D-printing technique called vat photopolymerization, which utilizes light to solidify extremely thin layers of liquid resin, fabricating a complex device one layer at a time.

This extremely precise process enabled the researchers to print layers that were only 25 micrometers tall, just a fraction of the width of a human hair. In this way, they could generate the complex internal geometry needed for a triaxial electrospray emitter.

Refining the design

The array, which is slightly larger than a U.S. penny, contains a network of internal coiled channels that carry liquid to 16 nozzles. These helical microchannels help maintain a uniform spray of microdroplets across all nozzles, while keeping the device as compact as possible.

“In a sense, the emitters in the array never learn they have company, or otherwise there would be cross-talking and causing interference between them. We achieved uniformity because of the work that went into our designs,” Velásquez-García says.

They also needed to fabricate extremely tiny channels without support structures, which could clog the device, and ensure all uncured resin was removed before the array was used.

The microchannels funnel liquid to the concentric nozzles, which must be perfectly aligned to properly emit microdroplets in a consistent manner.

“We were able to aggressively optimize the design because we could iterate in a much timelier manner. This ability to exquisitely refine designs is a key advantage of 3D printing,” Velásquez-García says.

The researchers tested multiple architectures to determine the ideal combination of liquid flow rates to maximize the stability and consistency of emitted microdroplets. They were surprised to find that the viscosity of the middle liquid plays the most important role in achieving stability in a microdroplet, since it preserves the thickness of each layer.

In addition, the researchers found that by adjusting flow rates and voltages, they could precisely tailor the thickness of each microdroplet layer. This would allow scientists to design drug-delivery particles with ideal layers so medicine releases at exactly the right time.

“By making such intricate devices more practical, we can empower others to pursue entrepreneurial and scientific advances,” Velásquez-García says.

In the future, the researchers want to continue refining their fabrication process and designs to achieve even smaller dimensions and integrate conductive or dielectric materials to the devices to make more advanced electrospray emitter arrays.

This research was funded, in part, by the Tecnológico de Monterrey – MIT Nanotechnology Program.

Innovative projects explore ways to deal with extreme heat

David L. Chandler — Mon, 08 Jun 2026 16:45:00 -0400

When MIT mechanical engineering Professor Kripa Varanasi landed in New Delhi in the middle of the night in June 2024 to attend a conference, he found himself in 104-degree Fahrenheit heat.

“This was June, and it was crazy. It was so hot for the whole meeting that I never left the hotel,” with daytime temperatures nearing 122 F.

It didn’t used to be that way. “When I grew up in India, it was not like this,” Varanasi says. “That kind of inspired me.”

He found a way to begin tackling the issue through a grant from the MIT Climate Project that provided seed funding to develop a proof-of-concept prototype of a wearable personal cooling system. The grant was one of four that were part of a Critical Cooling initiative for which the Climate Project requested proposals last year. The projects, which received grants totaling $450,000, are now complete. All have showed promise, and are now exploring ways to further develop their concepts.

Another MIT researcher, Yet-Ming Chiang, the Kyocera Professor of Materials Science and Engineering, looked into the potential of subsurface wells with heat-absorbing materials to supply spaces with air far below peak ambient temperatures while using much less energy than evaporation-compression heat pumps. The aim would be to use such systems in both small apartment buildings and single-family homes in India and other parts of the Global South.

Meanwhile, Asegun Henry, the George N. Hatsopoulos Professor in Thermodynamics, studied the use of an alternative approach to air conditioning to be more energy efficient and eliminate hydrofluorocarbon refrigerants that are potent greenhouse gases. His approach uses a cheap, widely abundant solid “caloric” material — rubber — to obtain a cooling effect, and then uses plain water as an efficient heat transfer fluid. The initial target market is single-family houses and apartment buildings, although larger systems could also serve data centers.

And Gang Chen, the Carl Richard Soderberg Professor of Power Engineering, addressed the tendency of existing air conditioning units being expensive and power hungry. They also use refrigerants that are far more potent greenhouse gases than carbon dioxide — and the coolants are likely to leak out when the devices are ultimately disposed of, adding to their global warming contribution. To help address that, Chen’s approach is to use a completely different kind of chemical refrigerant that has no greenhouse impact.

Christoph Reinhart, the Terri and Alan Spoon Professor of Architecture and Climate who leads MIT’s Sustainable Design Lab (SDL), championed the seed fund effort and served as faculty lead. “The term ‘critical cooling’ stems from a collaboration between SDL and Harvard’s Human Rights Entrepreneurs Clinic,” he says. “It is motivated by the fact that climate change increasingly causes heat fatalities, primarily among vulnerable populations, who lack access to active cooling. The impact that MIT can have by ‘cooling people, not spaces’ is enormous.” This vision led to the creation of the grant program, where each of the teams received funding for six months to see what they could do and explore really innovative approaches to the problem.

In collaboration with the Abdul Latif Jameel Poverty Action Lab (J-PAL), led on J-PAL’s side by Senior Policy Manager Andre Zollinger, the teams started with a workshop that brought together representatives from the World Bank, leaders from the Global South and industry, and engineers with ideas to suggest.

All of the teams made progress and most produced initial prototypes, says Liana Frey, a managing director at the MIT Climate Project, and an effort will be made to further develop and fund these ideas. “We’re continuing to look at different ways of proceeding with the work.”

One of these ways is through air conditioning. Worldwide, air conditioning is only available to about 8 percent of people — and that amount already contributes between 3 and 4 percent of global warming emissions — explains Chen. Meanwhile, the need for air conditioning and other ways of addressing extreme heat is steadily growing as the planet steadily warms up, and many of the people who will be most affected live in regions with limited access to reliable or affordable power and with high levels of poverty. The market for air conditioners is expected to triple or quadruple in coming years, he says, and their contribution to global warming will grow accordingly.

Chen says that he already had some ideas, but he hadn’t had a chance to test them out in experiments, which the grant enabled him to do. After building three prototypes and testing them out, he says, “I’m not at the stage where I can say that I know this will work.” But based on the experiments, he’d like to proceed to build a further prototype. If it works as well as expected, it would make a dramatic difference in air conditioning technology worldwide, including for the intensive cooling needs of new data centers.

Meanwhile, Varanasi’s way of looking at the problem was to consider individuals, not spaces. His devices work through the same principle as how an elephant uses its huge ears to dissipate heat and cool its blood.

The wearable device only consumes about 33 watts, he says, whereas a typical room air conditioner consumes around 1,000 watts. At U.S. material prices, the prototype device would cost about $20, he says, but if sourced with local material in India, he estimates it could be produced at a cost of less than $1 each.

Such garments could be bought in large quantities by the government and distributed to communities, where local entrepreneurs could set up charging stations to recharge the devices after a night’s wear, and other locals could set up businesses to manufacture the systems. The socks themselves would be washable, separately from the cooling material itself. This could enable people to at least get a good night’s sleep even in the extreme heat, he says.

The proof of concept he built used a simulated foot containing a heater, and measured the cooling effect. “We were able to keep it in the zone that we need for the body to stay cool,” he says. “So our initial prototype that we were able to build with this funding showed that this can become a viable solution.”

The same material could be used in other ways, such as to make sleeping bags with built-in cooling, he says. The raw material is widely available, but would be treated in a way that they developed. “It was a fundamental science bottleneck that we were able to overcome, which makes it possible.”

Varanasi says he is exploring various possibilities for how to develop his novel cooling material into a commercial product. “Ultimately, to make anything work, it has to be a business, otherwise good ideas can die,” he says. “It has to be a good business and a sustainable business.”

Luckily, there’s still support for advancing this work. “There are a lot of people interested in this heat-stress question,” says Frey. “It’s just becoming more and more urgent.”

MIT astronomers discover the earliest known flickering quasar

Jennifer Chu | MIT News — Mon, 08 Jun 2026 05:00:00 -0400

A supermassive black hole lies at the heart of every galaxy, including the Milky Way. When a black hole is active, it pulls material in as a whirlpool of high-temperature gas and dust. As this cosmic material piles up and falls onto a black hole, it lights up its vicinity, radiating a huge amount of energy.

The most energetic supermassive black holes are known as quasars, and they are some of the most active and luminous objects in the universe. These voracious systems take in so much material that the energy they emit can outshine all the light in the surrounding galaxy. The pattern of light from a quasar can give scientists clues to how active supermassive black holes shape the galaxies around them.

Now astronomers at MIT and elsewhere have detected a quasar flickering from the very early universe. The scientists traced the light from the quasar back to the “cosmic dawn,” just 850 million years after the Big Bang. The discovery represents the earliest flickering quasar detected to date.

“Although there have been a lot of quasars found in the cosmic dawn, this is the first time we actually see one flickering,” says Gene Leung, a postdoc in the MIT Kavli Institute for Astrophysics and Space Research.

The quasar’s flicker enabled the researchers to determine that, surprisingly, the ancient quasar’s whirlpool of gas and dust, known as an accretion disk, resembled a flat pancake, similar in shape to that of more modern-day quasars.

Their findings add to a longstanding mystery in cosmology: Why do supermassive black holes exist so early in the universe’s history? Physicists have assumed that a flat accretion disk reflects a relatively mature black hole that is in a calm and stable state. Black holes that are just starting to form, like those in the very early universe, should be more unsettled systems, with accretion disks that appear more puffy and chaotic.

The flat accretion disk around this very early quasar heightens the mystery of how supermassive black holes can grow and mature in a very short amount of cosmic time.

“I think what this suggests is that all the messy, very rapid growth phases that we expect all black holes to go through at some point happen very, very early on, before we see them as these very bright luminous quasars,” says Anna-Christina Eilers, assistant professor of physics at MIT. “That’s the picture that’s emerging.”

Eilers, Leung, and their colleagues report their results in a paper appearing today in Nature Astronomy. Their co-authors include members of MIT Kavli and multiple other institutions.

Past a pinprick

A supermassive black hole can be billions of times more massive than the sun. These gravitational giants are the central “engines” of most galaxies, helping to regulate a galaxy’s star formation and growth.

“Without supermassive black holes, no galaxy would look the way it does today,” Eilers says. “Black holes play a major role in shaping how galactic ecosystems look.”

It was long assumed that it should take more than a billion years for the first galaxies to settle and mature, so scientists didn’t expect to see supermassive black holes in the very early universe. But observations since the early 2000s showed otherwise. Scientists have spotted more than 200 supermassive black holes in the universe’s first billion years. Such objects were detectable because they were in an extremely active quasar phase, giving off enormous blasts of radiation that could be seen from Earth, 13 billion light years away.

These earliest quasars were observed as pinpricks of light, which signal the existence of a supermassive black hole at early times. But from these bright and distant dots, scientists aren’t able to tell much more about the black holes and their cosmic dawn environments. To do so, they need to catch a quasar’s “flicker.”

“People have known that quasars in the nearby universe can flicker,” Leung says. “The flickering comes from fluctuations in the way the gas is being fed into the black hole. And how a quasar flickers tells us something about the structure of a black hole’s accretion disk, and the kind of ‘bites’ that the black hole is eating.”

Mapping a flicker

Leung and Eilers looked to detect a flickering quasar from the early universe in hopes of learning more about the shape and structure of the earliest supermassive black holes. To do so would be a technical challenge: The further back in time and space an object is, the more distorted its light appears. This effect is due to the expanding universe, which effectively stretches, or “redshifts” light to redder, longer wavelengths. The same stretching occurs in time: Any flicker that naturally occurs over several weeks, for instance, would appear stretched out, flickering only every few months when seen from billions of light years away.

To spot a flickering quasar from the cosmic dawn, the team needed to observe the distant universe at redder wavelengths, and specifically within the infrared spectrum, and over long timescales of many years.

“This was the technical challenge we had to overcome,” Eilers says. “We needed data at longer, infrared wavelengths taken repeatedly over very long timescales.”

The team ultimately found a flicker in data collected by NASA’s Near-Earth Object Wide-field Infrared Survey Explorer (NEOWISE) mission — a space-based infrared telescope that scanned the entire sky over a total of about 14 years. Former MIT postdoc Kishalay De, who is now a faculty member at Columbia University, had launched a project to re-process archival data from NEOWISE. Based on the re-processed data, the team unearthed a signal, from just 850 million years after the Big Bang, which was confirmed to be the earliest flickering quasar.

“We saw the quasar flickering randomly over the 14-year period, much like a candle’s flame flickers without a fixed pattern,” Leung notes.

They estimate that the quasar is as bright as 12 trillion suns, and it is flickering by about 20 percent, meaning that it fluctuates up and down, by a brightness of about 2 trillion suns.

The researchers also tracked how the quasar’s light flickered over several different wavelengths. The wavelength of light reflects a certain temperature of the material that is emitting the light. The closer material is to a black hole, the hotter it is. Researchers can therefore use wavelengths of light to map the shape and structure of material within the accretion disk around a black hole.

Using NEOWISE data, the team analyzed the quasar’s flicker to determine the shape of the accretion disk surrounding the central supermassive black hole. They found that the disk is surprisingly thin and flat — a structure that astronomers mostly see around nearby, older black holes, that have had much longer to settle and mature.

“This provides direct evidence that the same feeding processes and structures observed in the nearby universe were already in place at very early times, despite very different cosmic environments, which had never been seen before,” Eilers says.

“This means something happened even earlier on that led to these systems to look so mature,” Leung adds.

The team hopes to peer even further back in cosmic time to catch a quasar’s earlier, premature development. Then, scientists can start to piece together the conditions that brewed up the first supermassive black holes.

This research was supported, in part, by NASA.

Improving the performance of high-power electronics

Adam Zewe | MIT News — Mon, 08 Jun 2026 00:00:00 -0400

The silicon that forms the foundation of most computer chips has fundamental limits to how much power it can manage, which constrains the speed and energy-efficiency of wireless communication systems.

A promising solution is to build future wireless electronics out of transistors made from gallium nitride, an advanced material that can handle the speed and energy required for demanding wireless applications like 6G and satellite communications.

But even in the best transistors, a very large fraction of that energy becomes heat. As researchers pack more gallium nitride transistors into a smaller area on a silicon chip, localized hot spots degrade reliability and hamper performance.

Now, a team from MIT and elsewhere has broken through this bottleneck by embedding gallium nitride transistors into an ultrathin layer of diamond. The diamond acts as a heat spreader that normalizes the temperature and allows the transistors to approach peak performance without reducing reliability.

The researchers used this technique to manufacture a power amplifier for wireless communications, which outperformed every similar amplifier they found in the literature.

While their fabrication technique is extremely precise and requires the integration of different material systems, it can be performed at the scale needed for commercial applications.

“No single material can do everything well in a wireless device, so these 3D heterogeneously integrated systems are here to stay. The key challenge left has been reliability and thermal management, and we might have now unlocked the final step we need to make these systems operate at scale and high volume,” says Pradyot Yadav, an electrical engineering and computer science (EECS) graduate student at MIT and lead author of a paper on this advance.

Yadav is joined on the paper by Tomás Palacios, the Clarence J. LeBel Professor of EECS, director of the Microsystems Technology Laboratories (MTL), and the MIT Institute for Soldier Nanotechnology; and Ruonan Han, a professor in EECS and a member of MTL and the Research Laboratory of Electronics; as well as others at Georgia Tech and Penn State University. The research was presented at the Radio Frequency Integrated Circuits Symposium, part of the IEEE International Microwave Symposium.

A multimaterial method

To build faster and more energy-efficient electronics, researchers are studying heterogeneously integrated systems in which multiple materials are stacked into a unified package to leverage the beneficial properties of each one.

For instance, MIT researchers previously stacked gallium nitride (GaN) on top of silicon as well as on top of glass to create higher-performance chips.

But in a heterogeneously integrated chip, each material has a different operating temperature, which can degrade the reliability of an electronic device.

“If we can incorporate a material that manages the heat so the GaN and silicon are at the same temperature, then the reliability of the entire 3D chip will improve. The best material for that is diamond,” Yadav explains.

The researchers use lab-grown, jewelry-grade diamond — the same type one would find in some engagement rings. Diamond has the highest thermal conductivity of any known material.

Advances in the growth process have significantly reduced the cost of single-crystal diamond wafers, making their use in computer chips more feasible.

In prior work, scientists have grown ultrathin, single-crystal layers of diamond on top of GaN transistors to manage heat.

But this growth process, which is not easy to scale up, introduces unwanted capacitances in the chip. These store energy flowing through the circuit, diverting it from the transistors and slowing down their operations.

The MIT researchers developed a completely different approach that reduces these unwanted capacitive effects. They embedded extremely tiny GaN transistors, known as dielets, into an ultrathin interposer, or substrate, made of single-crystal diamond. This diamond layer spreads and manages the heat, so the GaN and silicon operate at the same temperature without the unwanted capacitances.

“By putting these GaN transistors into a diamond interposer, we are actually able to improve the performance of the device, as opposed to degrading it. We can get the best of both worlds,” Yadav says.

Meticulous manufacturing

The fabrication process begins with the use of a lightning-fast femtosecond laser to cut prepared gallium nitride dielets out of a wafer.

The researchers use the laser to drill precisely sized cavities into the diamond substrate. They carefully place a die attach film, which is only 20 microns thick, at the bottom of the cavity and drop a dielet on top of the film.

Once the dielet is in place, they apply heat and pressure to mold it with the film and diamond substrate.

“That interface is key. If you don’t have that thermal die attach film placed just right, then the heat flow through the diamond to the GaN transistor will not be good enough. So you really need to have a very smooth, clean surface,” Yadav says.

The researchers then stack additional dielectric and metal layers on top of the GaN and diamond to build a working circuit.

They used this technique to fabricate a power amplifier, which is one of the key building blocks of any wireless system. Power amplifiers convert small electrical signals into larger ones that can then be transmitted long distances.

The amplifier they developed achieved higher output power, efficiency, and gain than any similar device the researchers are aware of, including an amplifier they designed in prior work.

“The power amplifier is the beating heart of a wireless device front end. Its performance will dictate the entire performance of your communication system. Our amplifier is powerful enough to ensure that a signal can be propagated for miles,” Yadav says.

These results show how their technique could be well-suited for demanding applications, like high-power radars, space communications, and industrial drones.

It could also be used to manage heat in systems that perform power conversions inside data centers, improving energy-efficiency.

Yadav hopes other researchers will build on these advances as they develop more complex heterogeneously integrated systems, opening the door to new possibilities with next-generation electronics.

“When I started my PhD, we wondered if any of this was even doable. It seemed like science fiction. Now we’ve shown all these systems that have outperformed anything that exists on the market today. GaN and 3D heterogeneous systems are going to be at the forefront of so many future applications. It is rewarding to know that we contributed a little bit to that space,” he says.

This research was funded, in part, by the Department of War, the Air Force Office of Scientific Research, the MIT Institute for Soldier Nanotechnologies, and the Qualcomm Innovation Fellowships. Device fabrication and microscopy were conducted at MIT.nano and the Georgia Tech Institute for Matter and Systems.

Startup helps retailers track their products in real-time

Zach Winn | MIT News — Fri, 05 Jun 2026 00:00:00 -0400

When you picture a worker at a retail store, you probably think of someone at a cash register or helping a customer. But employees also spend a lot of their time combing through stockrooms and shop floors, fulfilling requests or online orders and generally trying to keep track of all their inventory.

Keeping track of inventory takes so much time, in part, because retailers don’t always know where everything is located. That’s why when you ask a store associate to check if they have a shirt in your size, it may take them 20 minutes to get back to you.

Cartesian is helping retailers keep track of inventory with a technology invented at MIT. The system uses wireless signals from radio frequency identification (RFID) tags attached to items to find their precise location in a store, from the stockroom to the shop floor.

Last year, Cartesian did a study with a retailer and found its platform delivered meaningful annual savings at the store level by streamlining inventory tracking, optimizing workflows, and improving customer experiences.

“The big problem we’re solving is that about 50 percent of working hours in retail stores go to managing inventory,” says co-founder Fadel Adib SM ’13, PhD ’17, an associate professor at MIT. “That is roughly a $15 billion problem in the U.S. alone. We use algorithms to decipher indoor locations using wireless signals. The core technology enables a new level of indoor localization.”

Cartesian is already deployed in more than 700 stores across 15 countries and is working with one of the world’s largest fashion groups, Inditex, which is the parent company to brands like ZARA, Pull&Bear, and Oysho.

Beyond retailers and warehouses, Cartesian’s platform could also improve indoor location tracking for manufacturers, logistics operators, and robotics companies.

“The broad vision for what we are doing is spatial AI,” says Adib. “Today, AI does extremely well in the digital world. Now it has to move into the physical world. That means allowing machines to perceive their environment in such a way that they can interact with it. That’s where spatial AI comes in and where Cartesian sits.”

From technology to product

Adib, who holds a joint appointment in MIT’s Media Lab and Department of Electrical Engineering and Computer Science, has been studying wireless signals at the Institute for more than 15 years, dating back to research during his master’s degree.

“My group today researches how to use wireless signals to sense the world in ways that were not possible before,” Adib says. “We develop the fundamental technology and then we build systems around them. Our goal is to see these systems deployed in the real world for impact.”

When Adib joined MIT’s faculty, the first project he worked on was indoor localization using RFID tags. Isaac Perper ’20, MEnG ’21 later joined his lab as a student, and together they developed machine-learning algorithms to process RFID data to translate them into location patterns, with an initial focus on helping robots locate RFIDs indoors.

In 2021, Adib went through the National Science Foundation’s I-Corps program, which challenges researchers to interview potential customers to find the right problems to solve with their technologies. That’s when he realized how big of a problem inventory management is for retailers.

Cartesian was officially founded by Adib and Perper in the beginning of 2023, after they received a small business award from the National Science Foundation. The pair worked with MIT’s Technology Licensing Office to license patents from Adib’s lab. They also received support from MIT’s Venture Mentoring Service.

“Our goal was to reduce the cost of the technology to make it scalable,” Adib recalls. “Isaac focused on simplifying the product, leveraging progress in machine learning, and making it fast. It was a lot of iterating and testing early on.”

Retail workers spend much of their time locating items for a number of reasons. They might get an online order to fulfill, need to restock store shelves, or get a customer inquiry about items in the back.

Stores differ in how they organize their inventory. Most separate items by categories in specific shelves and bins then use barcodes or inventory systems that tend to get outdated fast.

“It’s a big problem for stores because customers may just leave before asking an employee to look for their size, or customers may get frustrated and leave if it takes too long,” Adib says. “The associate also wastes time looking for items they could spend doing higher-value work.”

Cartesian’s platform works with retailers’ existing handheld RFID readers, which store associates already use to manage inventory. Each store installs Cartesian’s software into their existing inventory apps or uses a custom app for employees to access directly.

“The RFID readers are how stores tell what’s in stock and what’s out of stock,” Perper says. “We figured out a way to leverage the same scans they’re already using with the reader, put the data they generate into our machine-learning algorithms, and generate maps of where all the items are.”

Customers can build analytics on top of Cartesian’s technology to keep track of inventory levels, show customers maps of where each item is located, and create other services.

“They use our location intelligence platform and build different products on top,” Adib says. “We can work with any device, any store, any type of RFID. It’s a simple interface. All the sophisticated location algorithms sit in the cloud.”

Beyond retail

Cartesian signed its first big contract in 2025 and soon expanded to several hundred stores. One of Cartesian’s advantages is its ability to quickly scale. Perper says they can add a store in about one minute. Cartesian’s team doesn’t even have to travel to a new store to turn on its system if it’s already working with the company.

“It’s as simple as flipping a switch, preparing the data, and sending it to our customers,” Perper says. “One of our first big bets was, ‘Can we build this entirely on existing hardware?’ That bet is starting to pay off.”

Cartesian’s models can also work with Wi-Fi and Bluetooth signals, which the company plans to use with customers in other verticals.

“Right now, we’re focused on applications in retail, but this technology has a lot of value in manufacturing, warehouses, and other locations,” Adib says.

Cartesian’s team aims to be deployed in tens of thousands of stores over the next year and then begin expanding beyond retail into industries like manufacturing and robotics.

“What’s most exciting about Cartesian to me is we’ve built a lot of the technology foundation, and now that we have the fundamentals in place, we hope to build specific application layers,” Perper says. “Then we can ask customers in different verticals about their problems and apply our technology in different ways to solve it.”

Developing innovative alternatives to conventional carbon capture methods

Molly Chase | Climate and Sustainability Consortium — Thu, 04 Jun 2026 16:35:00 -0400

Carbon capture is an important climate change mitigation strategy, but it faces technological barriers and can be energy-intensive and expensive. To help make necessary advances in this area, a team of MIT researchers, with support from the MIT Climate and Sustainability Consortium (MCSC), are exploring energy-efficient and scalable alternatives to conventional carbon dioxide (CO₂) capture methods.

Conventional amine scrubbing, which is the current standard for CO₂ capture, is energy-intensive and difficult to scale, limiting its impact despite the urgent need to reduce carbon emissions and upgrade CO₂ into valuable products. In a new article published in Nature Energy, MIT researchers — graduate students Fang-Yu Kuo of the Department of Chemical Engineering, and Gi Hyun Byun of the Department of Mechanical Engineering (MechE); Professor Betar Gallant of MechE; and former MCSC postdoctoral Impact Fellows Glen Junor and Akachukwu Obi — investigate a promising alternative to these conventional CO₂ capture methods. Their findings could move the needle on achieving efficient and flexible carbon capture and removal.

In their paper, the team explores an alternative, electrochemically mediated CO₂ capture (EMCC). This approach enables electrification of CO₂separation — driven ideally by renewables — but currently faces challenges, such as relying on sorbents that require highly reducing potentials, where oxygen reduction side reactions become significant. This can compromise both efficiency and long-term performance. To tackle this shortcoming of EMCC, the MIT team researched whether N-heterocyclic imines (NHIs) is a useful new class of EMCC sorbent.

“NHIs have shown promise in recent years as CO₂ sorbents because of the ease of NHI molecular modifications for tuning basicity,” says Fang-Yu Kuo. “Our work translates these NHIs for the first time into the EMCC application space, and demonstrates that NHI-based sorbents can be modulated electrochemically for CO₂ separation by a unique separation mechanism that avoids the need of applying highly reducing potentials.”

The team’s initial research establishes a novel bis(NHI) structure that can enable a theoretical CO₂ modulation of two molecules per electron during cell operation. The initial published result also indicates that with further molecular engineering of bis(NHI) structures to strengthen CO₂ binding affinity, the bis(NHI) could operate in more diverse electrolyte environments, opening new possibilities to optimize system performance in terms of electron efficiency, energy efficiency, and operational flexibility.

“A critical future direction of our work involves gaining deeper mechanistic insight into the stability and degradation pathways of the bis(NHI) radical cation,” says Kuo. “Understanding these pathways will inform the rational design of next-generation bis(NHI) molecules, enabling longer operational lifetimes and enhanced cycling durability for practical deployment.”

NSF renews support for MIT-led AI and physics institute, expanding a new model for discovery

Laboratory for Nuclear Science — Thu, 04 Jun 2026 12:00:00 -0400

The MIT-led Institute for Artificial Intelligence and Fundamental Interactions (IAIFI) has received renewed support from the National Science Foundation (NSF) for an additional five years, increasing annual funding from $4 million to $4.98 million. The renewal marks a new phase for IAIFI, which has spent its first five years building a research model and an interdisciplinary community around a central premise: that AI can open new ways of doing physics, while physics can help mold better AI systems.

Launched in 2020 as part of the National Artificial Intelligence Research Institutes program, IAIFI brings together researchers from MIT, along with Harvard, Northeastern, Tufts, and Boston universities. Its work has shown that machine learning can accelerate discovery in physics, while insights from physics can make AI systems more principled and interpretable.

“From the beginning, IAIFI has been built around a two-way street: AI enabling better physics, and physics enabling better AI,” says Jesse Thaler, IAIFI’s director and a professor of physics at MIT. “We have seen this virtuous cycle play out across multiple areas of physics and AI over the past five years. The exchange is producing not just new results, but genuinely new ways of doing science.”

Research across physics and AI

IAIFI’s research spans particle physics, nuclear physics, astrophysics, and foundational AI, with many advances emerging from collaborations across those areas.

In particle physics, IAIFI researchers have developed AI techniques to handle the immense data rates from the Large Hadron Collider in real-time, helping turn a firehose of collision data into actionable physics. In nuclear physics, IAIFI researchers are using AI-based generative methods to model the interactions of quarks and gluons in lattice quantum chromodynamics, creating new ways to study the structure of matter from first principles. In astrophysics, machine learning is being used to uncover new cosmic phenomena and improve the sensitivity of the MIT-led LIGO gravitational-wave experiment.

At the same time, ideas from physics are informing the development of new AI methods. IAIFI researchers are developing learning algorithms and new model architectures that embed physics knowledge and best practices — including symmetries, geometric structures, exactness guarantees, and statistical methodologies — directly into neural networks, producing systems that are more reliable, interpretable, and data-efficient.

“AI has begun to transform how physicists tackle some of the field’s most challenging problems,” says Mike Williams, interim director of IAIFI and a professor of physics at MIT. “More importantly, it is starting to expand the frontier of what problems we can realistically address, making it possible to pursue questions that were once completely beyond our reach.”

Training the next generation

A defining feature of IAIFI is its investment in people. The IAIFI Postdoctoral Fellows program supports early-career scientists pursuing research at the intersection of physics and AI, pairing each fellow with mentors in both domains and fostering collaboration across institutions.

Eight fellows have completed the program to date. Three have secured faculty positions; others have taken research roles at leading AI companies or joined startups, reflecting how broadly the skills cultivated at IAIFI translate.

“The IAIFI Fellowship shows what can happen when early-career scientists are given the freedom and support to work across traditional boundaries,” says Phiala Shanahan, IAIFI’s interim deputy director and a professor of physics at MIT. “Our fellows aren’t just contributing to physics or to AI separately — they are helping shape a growing field at the intersection.”

IAIFI’s annual PhD Summer School has become a focal point for the growing community of “centaur scientists” with expertise in both physics and AI. For the 2026 edition, the program received nearly 600 applications for roughly 100 in-person spots, with about 300 additional participants expected to join virtually. Previous participants have strongly recommended the school to their peers for its combination of lectures, hands-on tutorials, coding sprints, and networking events.

At MIT, IAIFI has helped shape new educational pathways, including an interdisciplinary PhD program in physics, statistics, and data science — a collaboration between the Department of Physics and the Statistics and Data Science Center — which has awarded 20 doctoral degrees since 2021. IAIFI members Phil Harris and Isaac Chuang have also developed a course on computational data science in physics, offered both on campus (Course 8.16) and as a free online course through MITx.

A growing community

Beyond its core research and training programs, IAIFI convenes researchers through its annual summer workshop, which will be held this year at the MIT Schwarzman College of Computing building. The institute also engages the broader public through collaborations with the MIT Museum, the Museum of Science in Boston, hackathons, and widely viewed online content exploring AI and physics.

“IAIFI shows what becomes possible when researchers in physics, computation, statistics, and data science organize around shared scientific questions,” says Nergis Mavalvala, dean of the MIT School of Science and the Curtis and Kathleen Marble Professor of Astrophysics. “That kind of sustained, cross-disciplinary collaboration is essential to the future of scientific discovery.”

IAIFI is hosted in the Laboratory of Nuclear Science at MIT, led by Director Jesse Thaler (currently on sabbatical), Interim Director Mike Williams, Interim Deputy Director Phiala Shanahan, and Managing Director Marisa LaFleur, along with steering committee members Lisa Barsotti, Isaac Chuang, Will Detmold, Bill Freeman, Phil Harris, Lina Necib, Tess Smidt, and Marin Soljacic (and steering committee members from other IAIFI universities).

Looking ahead

As a member of the National Artificial Intelligence Research Institutes program, IAIFI is part of a nationwide effort to advance AI-driven discovery and innovation.

“The connections among the NSF AI Institutes have been as valuable as the work within them and continue to grow,” says Marisa LaFleur, IAIFI's managing director. “We’re sharing management strategies and resources for training, community building, and collaboration that make the whole network stronger.”

For IAIFI, the renewed funding is an opportunity to push deeper into what the institute calls the “physics of AI” — using physical reasoning, physical challenges, and physical tools not just to apply AI, but to understand and improve it. That agenda, along with a growing community of researchers trained to work across disciplines, is what drives the institute's next phase.

“The first phase of IAIFI established the model: interdisciplinary research, early-career talent, and a dynamic community, organized around the idea that AI and physics make each other stronger,” Thaler says. “Now we have the foundation — and the entrepreneurial spirit of our centaur scientists — to push that model into new territory and raise our ambitions.”

Teaching AI agents to ask better questions by playing “Battleship”

Alex Shipps | MIT CSAIL — Wed, 03 Jun 2026 17:00:00 -0400

In 2026, the hype for artificial intelligence agents is louder than ever before. These semi-autonomous programs can “think” and execute well-defined tasks in areas like customer service and software development, typically using language models (LMs). But fields like medical diagnosis and scientific discovery require them to inquire about a vast range of solutions in uncertain environments, which LMs struggle with.

Researchers at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) and Harvard University’s School of Engineering and Applied Sciences (SEAS) peered deeper into LMs to understand their main issues in high-stakes settings. Their test: “Battleship,” a classic guessing game that’s helped cognitive scientists study how humans seek information.

CSAIL and SEAS scholars added a twist by reframing the game around asking and answering natural language questions. In their “Collaborative Battleship” game, one participant is a “captain” who inquires about where hidden ships are, while their teammate plays the “spotter” by responding to those questions in real-time.

The researchers first had over 40 humans play the game together, collecting their questions and yes-no answers to build the “BattleshipQA” dataset. These results were a helpful point of comparison when the team tested state-of-the-art LMs (like GPT-5) and smaller models (like Llama 4 Scout) on their game. Without training the models beforehand, they found that top LMs can “beat” humans at “Battleship” — that is, complete the game in fewer turns — but smaller systems are far less rational.

The chief issue was that many models are simply not adept at coming up with useful questions. To get LMs to inquire in ways that reveal more information about hidden ships, the researchers gave each model a Monte Carlo inference strategy, which carefully measures the likelihood of different options being correct with each response. The result: AI models that can beat regular players at “Battleship,” regardless of scale.

Perhaps the most striking results were Llama 4 Scout’s gains. As a relatively small LM, it only beat humans 8 percent of the time. But with refinements to its inference strategy, the model reached a “Battleship” win rate of 82 percent versus humans. This careful and efficient style of asking questions also enabled the model to outpace a frontier model (GPT-5), while operating at around 1 percent of its cost.

On top of this improvement, the researchers shrank the gap between humans and LMs in answering questions. While GPT-5 was a reliable spotter that helped models finish games faster, smaller systems had a bad habit of giving the wrong answers about where ships were hidden. The models saw an accuracy boost of 15 percent on average when they began converting questions into code that explicitly tells them how to verify their answers (for example, having the model run a quick search of an area when asked if a ship was there).

“Today’s language models are primarily optimized to answer complex queries, but it’s less clear whether they learn to ask good questions for themselves,” says MIT PhD student and CSAIL researcher Gabriel Grand SM ’23, who is a lead author on a paper about the work. “Our work shows that asking informative questions depends on the ability to predict and simulate the world. We find that when we give agents access to a ‘world model,’ they ask better questions and make discoveries more efficiently.”

A sea change for LMs

The team’s first focus was getting LMs to ask better questions. By implementing Monte Carlo inference strategies, the LMs reason about potential guesses as individual particles. The ones that appear more valid with each answer from the spotter would be weighted more heavily, sort of like game balls that inflate or deflate each turn. With this more calculated, adaptive approach, the captain could make inquiries that extracted considerably more info from the spotter.

The scientists then turned to the widely used programming language Python to help out AI spotters. Each question the captain asked was automatically converted into an encoded command. For example, a question like, “Is there a ship in column one that spans two rows?” turns into instructions for the spotter LM to search the area in question and assess how wide the digital game piece is. By giving the model clear directions in a language it understands particularly well, each system gave correct answers considerably more often. The lightweight system GPT-4o-mini saw a nearly 30 percent performance bump, for instance, and even the large model Claude 4 Opus jumped about eight points.

“The field has seen a lot of success from ‘auto-formalization’ strategies, in which LMs generate code to verify their solutions,” says senior author Jacob Andreas, an MIT electrical engineering and computer science associate professor and CSAIL principal investigator. “What I find most exciting about this work is that it opens up the possibility of using these techniques to generate better solutions in the first place, by improving LMs’ exploration and information gathering capabilities. We are excited to scale this work up from scientific domains to applications like coding and mathematical problem-solving.”

Let’s play something else

But how would this approach fare in other board games? The team tested their newly equipped LMs at “Guess Who?”, where large and small models skillfully whittled down 100 options to correctly guess which hidden character had been chosen. Llama 4 Scout was successful 30 percent of the time, but after Grand and his colleagues’ tweaks, it completed the task on over 72 percent of its runs. Meanwhile, GPT-4o leapt from 62 percent to 90 percent. GPT-5 was the spotter in each game to ensure questions were answered as accurately as possible.

While LMs have made promising progress in both games, there’s room for improvement. For instance, the models still struggle to answer complex questions, compared to humans. OpenAI researcher, recent Harvard graduate, and coauthor Valerio Pepe adds that “GPT-5 can beat your average ‘Battleship’ player, and gets a hair better with our methods. However, expert players are still hard to beat for all models, unlike in chess, where even top players don’t succeed against AI systems.”

The researchers’ findings show that AI agents have untapped potential in “needle-in-a-haystack” discovery — navigating a massive space of options to find a rare solution to scientific challenges. While improved information-seeking skills would make them excellent research assistants with, say, identifying a compound’s molecular structure, the researchers caution that “Collaborative Battleship” is a somewhat simple test bed. They’d like to test LMs in more complex settings, where the systems have to consider far more options.

Grand also plans to have humans and AI models collaborate to study whether they work better together. The models might also benefit from a bit of fine-tuning on game simulations, and with more computing power, LMs would have more advanced inference capabilities to predict how a game will evolve.

“As AI systems become more agentic, the hardest problems turn out to be social ones: tracking common ground, resolving misunderstandings, and adapting to different partners over time,” says Robert Hawkins, assistant professor of linguistics at Stanford University, who wasn’t involved in the paper. “This work elegantly captures these phenomena in a controlled collaborative setting, and makes a compelling case that the real bottleneck for AI agents isn’t just the calculation of optimal questions, but the pragmatic reasoning needed to make the most of their answers.”

Grand and Pepe wrote the paper with two CSAIL principal investigators: MIT Associate Professor Jacob Andreas and MIT Professor Joshua Tenenbaum. Their work was supported, in part, by the MIT Siegel Family Quest for Intelligence, the MIT-IBM Watson AI Lab, the FinTechAI@CSAIL initiative, a Sloan Research Fellowship, Intel, the Air Force Office of Scientific Research, the Defense Advanced Research Projects Agency, the Office of Naval Research, and the National Science Foundation. They showcased their paper as an oral presentation at the International Conference on Learning Representations (ICLR) in April.

A new vaccine adjuvant could make it easier to eradicate polio

Anne Trafton | MIT News — Wed, 03 Jun 2026 14:00:00 -0400

In the United States, children routinely receive an injectable form of the polio vaccine. This vaccine is very effective at preventing illness, but it doesn’t block transmission of the polio virus as well as the oral polio vaccine does.

Poliovirus is usually transmitted through contaminated food or water, so the GI tract is where the body is first exposed. Because the oral vaccine induces a mucosal immune response within the GI tract, it is much more effective at preventing infection and spread of the virus. However, there is a small chance that the oral vaccine can become infectious, so many countries have stopped using it.

Researchers at MIT have now come up with a way to modify the injectable vaccine so that it can also promote a mucosal immune response. This vaccine could help to achieve polio eradication while avoiding the risks of the oral polio vaccine.

“People who are vaccinated with the injectable vaccine are not getting sick, but they may be helping the virus circulate. Mucosal immunity could help lower that shedding and ideally eliminate it,” says Ana Jaklenec, a principal investigator in MIT’s Koch Institute for Integrative Cancer Research.

The researchers’ new vaccine consists of the current injectable, inactivated polio vaccine (IPV), delivered with a nanoparticle-based adjuvant that helps steer immune cells to the mucosal lining of the intestine. In a study of rats, the researchers found that this vaccine produced a 20-fold increase in the type of antibodies needed for mucosal immunity, compared to IPV alone.

Jaklenec and Robert Langer, the David H. Koch Institute Professor at MIT, are the senior authors of the study, which appears today in Science Advances. MIT postdoc Behnaz Eshaghi is the lead author of the paper.

Targeting polio

Polio, which can cause paralysis in severe cases, is now rare in most of the world due to extensive vaccination campaigns. The virus is highly contagious and is most commonly spread through consumption of food or water contaminated with the stool of an infected person.

Cases are occasionally seen in the United States and other countries, and the virus is endemic in Pakistan and Afghanistan. While most of these cases are caused by the virus spreading among unvaccinated individuals, some cases may be due to the evolution of the live viruses used in the oral polio vaccine (OPV). These viruses are attenuated, meaning they are alive but weakened. In rare cases, they can mutate and evolve to become infectious again.

It’s also possible that wild poliovirus can be spread by people who have received the injected polio vaccine. These people would likely not experience any symptoms, but they could still shed the virus in their stool. Eventually, this could expose someone who isn’t vaccinated. Studies have shown that even in countries that with very high polio vaccination rates, the virus can be detected in wastewater.

To boost the chances of completely eradicating polio, it would be ideal to use a vaccine that cannot evolve to cause infection, like the current injectable IPV, and would also induce mucosal immunity, like the OPV.

In hopes of achieving that, the MIT researchers teamed up with researchers at Harvard Medical School who have shown that using a derivative of vitamin A as a vaccine adjuvant can help stimulate immune cells to go to the GI tract.

That adjuvant, known as Am80, works well, but to generate a strong response, it needs to be injected for several days in a row, which is not feasible for most vaccine campaigns.

To eliminate the need for repeated daily injections, the researchers set out to develop a nanoparticle formulation that would enable the adjuvant to be released slowly over several days. They tested several different types of nanoparticles and found that the one that worked best was a lipid nanoparticle (LNP).

“The purpose of the nanoparticle is making sure that we can engineer a platform with a sustained release of the cargo for a few days,” Eshaghi says. “That way we can overcome the bottleneck that for free administration of Am80 you need multiple daily injections.”

Mucosal immunity

In tests in rats, the researchers delivered an injection of an inactivated polio vaccine, similar to the one that is now used in the United States, along with a separate injection of Am80 encapsulated in LNPs. After the first dose, boosters were given at four weeks and eight weeks.

After injection, the nanoparticles accumulate in the lymph nodes, where they interact with B and T cells that are also exposed to the polio vaccine. This interaction stimulates the B and T cells to produce two surface proteins that act as homing signals directing them to the GI tract.

The B cells also begin producing a type of antibodies called IgA, which protect body surfaces from infection by coating the mucosal membranes. In addition, the rats also produce IgG antibodies that circulate in the bloodstream, similar to the antibodies that are normally produced in response to the injected polio vaccine.

“IPV is a safe vaccine, but it cannot create mucosal immunity. OPV can create that mucosal response, but it is not as safe,” Eshaghi says. “By adding Am80 to lipid nanoparticle as an adjuvant, we are combining the safety of IPV with an adjuvant that can produce the mucosal immunity that normally you can only get with OPV.”

The researchers now plan to test the vaccine in additional larger animal models, where they will inject the vaccine and adjuvant mixed together.

Using Am80 or other adjuvants to induce a mucosal response could also help researchers design improved vaccines for other pathogens that infect the GI tract, or for diseases that infect the lungs or reproductive tract.

“You could potentially add it to any vaccine that’s injected,” Jaklenec says. “This particular work shows that cells can be directed to the gut and increase enteric mucosal immunity. Whether it works for the respiratory or vaginal mucosa is not yet clear.”

The research was funded by the Gates Foundation.

MIT chemists design impact-resistant plastics

Anne Trafton | MIT News — Wed, 03 Jun 2026 11:00:00 -0400

With help from a novel cross-linking molecule, MIT chemists have shown they can substantially improve the ballistic impact resistance of common polymers, including polystyrene and a type of rubber used to make shoe soles.

Polystyrene is a hard, glassy polymer that is used to make many types of plastic containers, such as bottles and mugs, as well as disposable cutlery. It is also found in coatings for electronic devices, and its foam form is the basis for Styrofoam and other lightweight packaging. (While sometimes labeled with recycling code No. 6, polystyrene is difficult to recycle and rarely collected for reuse in the U.S.)

To make the polymer more resistant to sudden impact, the MIT team added weak bonds scattered throughout the material as cross-links, which allows the material to dissipate energy much more effectively under deformations. When struck by a projectile, these weak bonds selectively break at the site of impact to open up pathways for enhanced energy absorption.

The researchers found that this approach can also fortify styrene-butadiene-styrene rubber, and they are now investigating whether it will also work for other types of polymers such as latex or the rubber that is used to make tires.

“These cross-linkers can substantially increase the amount of energy that the material absorbs under ballistic impact. You can imagine many applications of that, especially if this could be generalized to other polymers,”says Jeremiah Johnson, the A. Thomas Geurtin Professor of Chemistry at MIT and a member of the Koch Institute for Integrative Cancer Research.

Johnson and Keith Nelson, the Haslam and Dewey Professor of Chemistry, are the senior authors of the study, which appears today in Nature. Former MIT postdocs Zhen Sang and Suong T. Nguyen and MIT graduate student Kwangwook Ko are the paper’s lead authors.

Tougher plastics

In a study published in 2023, Johnson and colleagues at MIT and Duke University showed that they could make polymers tougher using a counterintuitive strategy: adding weak cross-linkers that are distributed throughout a polymer network. These weak linkages, also called mechanophores, break under tearing conditions in a way that helps preserve the stronger bonds that bear the load, allowing the material to dissipate more energy.

“As a crack starts to propagate through the material, these mechanophores split in two, which helps to dissipate energy and redirect where the crack goes. That means you have to put in more energy to tear the material,” Johnson says.

Unlike their previous study, which examined toughening under slow tearing conditions, the new Nature study aimed to develop mechanophore-enabled strategies for resisting rapid deformation, such as that caused by sudden impact. The researchers were especially interested in applying the strategy to some of the most widely used polymers, such as polystyrene.

To do that, they developed a way to directly incorporate mechanophores as cross-links into common polymers. Then, they used a system invented by Nelson — laser-induced microprojectile impact testing (LIPIT) — to study how the resulting polymers respond to projectile impacts. With this system, tiny projectiles — silica beads about 10 microns in diameter — are fired at the film at about 750 meters per second (more than 1,600 miles per hour). The amount of energy absorbed by the material can be calculated by measuring the change in the particle’s velocity before and after it passes through the film.

“We first developed this method to study microparticle impact and penetration into bulk polymer samples, where we would monitor particle propagation through about 100 microns of material and analyze after impact how polymer morphology had changed,” Nelson says. “Our new measurements show how much additional information can be extracted from particle velocities before and after penetration through a thin layer. They also show deeply informative deformation patterns both during particle impact and afterward.”

This technique allowed the researchers to mimic the type of forces that might be seen in the real world when a plastic object is hit with another object, or when you drop your phone on the ground. In their experiments, the researchers showed that mechanophore cross-linked polystyrene was able to absorb substantially more energy from an impact than regular polystyrene.

“It turned out that the mechanophore leads to substantial increases in energy dissipation compared to both uncross-linked and conventionally cross-linked polystyrene, a behavior that had not been observed in related previous work,” Johnson says.

Absorbing impact

To figure out how the mechanophores help make polystyrene more impact resistant, the MIT team enlisted help from collaborators at MIT, Purdue University, Northwestern University, and Duke University.

Through experiments and simulations, they found that when a high-speed particle strikes the material, it raises the temperature at the impact site high enough to form a mobile zone. In this zone, the mechanophore bonds are selectively broken under force, opening controlled pathways that better absorb the energy of impact while leaving the area beyond the impact site relatively unaffected and stable.

“What is particularly attractive about this approach is the ability to bestow these properties upon ‘off-the-shelf’ commodity plastics, both glassy and elastomeric, with minimal chemistry which makes it in principle quite scalable and relevant. This study combines an elegant approach while providing an in-depth mechanical analysis of the failure mechanism,” says Yoan Simon, an associate professor in the School of Molecular Sciences at Arizona State University, who was not involved in the research.

The researchers also found that they could insert these mechanophores into styrene-butadiene-styrene (SBS) rubber — which is used in shoe soles as well as asphalt and roofing materials — and observe a similar effect. They are now exploring whether this approach could also work with a related material, styrene-butadiene rubber, which is one of the major components of tires.

If successful, this technology could yield longer-lasting tires and also cut down on the amount of microplastics generated when tires contact the road, which is estimated to account for at least 10 percent of the microplastics in the environment.

“Materials with energy-absorbing mechanophores could one day help keep your vehicle's tires from blowing out on the highway or provide more protective cases for personal electronics,” says Katharine Covert, program director of the U.S. National Science Foundation Centers for Chemical Innovation, which invested in the team’s research. “This work really demonstrates how valuable new insights can be rapidly generated by bringing together researchers with different areas of expertise.”

The research was funded by the National Science Foundation Center for the Chemistry of Molecularly Optimized Networks, the U.S. Army Research Office through MIT’s Institute for Soldier Nanotechnologies, a Schmidt Science Postdoctoral Fellowship, and the U.S. Air Force Office of Scientific Research.

MIT researchers teach AI models to interpret charts

Adam Zewe | MIT News — Wed, 03 Jun 2026 00:00:00 -0400

To accelerate and refine decision-making in a fast-paced, global marketplace, enterprises may deploy generative artificial intelligence models to help summarize and interpret the charts that often fill market summaries and financial reports.

But even the latest vision-language models sometimes struggle with this task, since it requires a model to integrate visual, numerical, and linguistic understanding. A company that invests in a state-of-the-art model might still receive inaccurate or incomplete information.

To fill this performance gap, researchers from MIT and the MIT-IBM Computing Research Lab developed a multifaceted resource for AI users that is specifically designed to teach vision-language models (VLMs) how to effectively interpret charts.

They used a novel data generation method to build a state-of-the-art dataset that includes more than a million varied charts. The dataset also encodes many visual, linguistic, and numerical components of each chart image, which enable models to robustly reason about the information in a chart.

The researchers used this dataset, called ChartNet, to train a series of open-source VLMs. Many of these smaller models significantly outperformed orders of magnitude larger, commercial models on tasks like data extraction and chart summarization.

By enabling open-source models to outperform their commercial counterparts, ChartNet could allow small firms with limited budgets to more readily utilize AI. The open-source dataset can be used to improve the capabilities of AI models for tasks like business trend analysis and scientific figure interpretation.

“We developed ChartNet to be a one-stop shop for chart understanding, covering basically anything that an AI model and a practitioner who is training that model might need. We hope our work motivates researchers to achieve state-of-the-art performance with smaller models that don’t require infinite amounts of computation,” says Jovana Kondic, an MIT electrical engineering and computer science (EECS) graduate student and lead author of a paper on ChartNet.

She is joined on the paper by many co-authors from MIT, the MIT-IBM Computing Research Lab, and IBM Research, including Pengyuan Li, a research staff member at IBM Research; Dhiraj Joshi, a senior scientist at IBM Research; Isaac Sanchez, a software engineer at IBM Research; Aude Oliva, director of strategic industry engagement at the MIT Schwarzman College of Computing, MIT director of the MIT-IBM Computing Research Lab, and a senior research scientist in the Computer Science and Artificial Intelligence Laboratory (CSAIL); and Rogerio Feris, a principal scientist and manager at the MIT-IBM Computing Research Lab. The research will be presented at IEEE Computer Vision and Pattern Recognition Conference.

A dataset bottleneck

Researchers have made great strides developing generative AI models that excel at natural language processing and reasoning about natural images. But less work has focused on interpreting complex multimodal data contained within charts, Kondic says.

Yet for large and small businesses in nearly every industry, chart understanding is a critical task.

“The finance industry thrives on charts. If vision-language models can extract information out of charts, like descriptions of trends, that facilitates a lot of workflows that happen downstream,” Joshi says.

The lack of high-quality training data is a major bottleneck holding back the development of VLMs that can accurately interpret charts. Many datasets contain limited chart images pulled from the internet and often lack the necessary scale and additional information to help a model interpret the underlying data.

“A vision-language model, unlike our brains, may need to see thousands of examples during training to reliably recognize something as a line chart,” Kondic says.

The researchers sought to overcome those shortcomings by generating synthetic data. Synthetic data are artificially generated by algorithms to mimic the statistical properties of actual data.

The ChartNet dataset holds more a million high-quality chart images, along with the corresponding code used to generate each chart, a textual description, and a table that contains its numerical information. In addition, each datapoint includes question-and-answer pairs to teach the model how to correctly answer questions about the chart image.

“These additional modes of data guide the model to connect and align the different pieces of information that the chart image encodes,” Kondic says.

Data generation

To build ChartNet, the researchers created a two-step, synthetic data generation pipeline.

First, their automated system translates any pre-existing set of chart images into code. Then the system iteratively augments that code to change different aspects of each chart, such as chart type, data values, topic, colors, etc.

“We can start from a single chart that we use as a seed and come up with hundreds of augmentations of it. This is how we were able to build a dataset with more than a million diverse images,” Kondic explains.

They also incorporated an automated quality check process to ensure the synthetic data are high quality. This process verifies that the code is executable and rendered chart images are accurate and clean.

“We don’t want to just be generating diverse samples. We also want the information to be presented in a meaningful way,” she says.

ChartNet also includes a selection of chart datapoints annotated by human experts. This provides access to additional types of charts and supporting data that carry validity guarantees.

A practitioner could use the annotated data to fine-tune an existing VLM, further boosting performance for a specific application, Joshi adds.

The researchers tested ChartNet by training IBM’s Granite Vision series of models as well as several other open-source models of various sizes and evaluating them on various chart interpretation tasks. The dataset improved the accuracy of all models in chart reconstruction, chart data extraction, chart summarization, and chart question answering.

With ChartNet, small open-source models consistently outperformed much larger commercial models.

“A lot of prior training datasets only focused on answering simple questions about a chart. We tried to go beyond that with ChartNet by generating data that support all aspects of robust chart understanding,” Kondic says.

In the future, the researchers plan to continue expanding ChartNet by incorporating data with added levels of complexity. They also want to draw on feedback from the research community.

This research was funded, in part, by the MIT-IBM Computing Research Lab.

Ultrasound-based pacemaker noninvasively steadies the heart

Jennifer Chu | MIT News — Tue, 02 Jun 2026 05:00:00 -0400

MIT engineers have developed a noninvasive pacemaker that stimulates the heart using ultrasound. The design could one day provide a surgery-free alternative to traditional cardiac implants.

The new device is designed as a small sticker that can be worn on the chest. Tiny transducers on the sticker send ultrasound pulses through the chest to stimulate the heart. The ultrasound waves trigger the opening of certain ion channels in heart cells, an effect the researchers amplified through genetic engineering. When the channels open, they let in calcium, which signals a heart cell to squeeze and beat.

In experiments in the lab, the researchers applied ultrasound waves to engineered human cardiac cells and found that the pulses effectively maintained the cells’ healthy contractions. They also tested the ultrasound sticker on rats and found the device quickly, safely, and noninvasively corrected arrhythmias and restored normal, regular heart contractions.

The team has fabricated a prototype that includes the ultrasound sticker (about the size of a postage stamp) and a small, pocket-sized device containing associated batteries and electronics. The same group previously demonstrated a sticker design that uses ultrasound to image deep organs and tissues. They now plan to combine the two approaches into one ultrasound sticker to simultaneously monitor and regulate the heart’s activity.

“We believe you could one day have stickers on the body that could do long-term imaging deep in the body and also do stimulation for therapeutic effects, in a noninvasive closed-loop way,” says Xuanhe Zhao, professor of mechanical engineering and of civil and environmental engineering at MIT.

Zhao and his colleagues, together with collaborators from Professor Qifa Zhou’s group at the University of Southern California (USC), have published their results in a study appearing today in the journal Nature Biomedical Engineering. The study’s MIT co-authors include first author Chen Gong, together with Runze Li, Won Jun Song, and former postdocs Gengxi Lu, Shucong Li, and Hsiao-Chuan Liu. Other collaborators include researchers from Harvard University, the University of California at Los Angeles, and other groups at USC.

Sound genes

Today, around 3 million adults in the United States live with pacemakers. The small battery-powered devices are surgically implanted in a person’s chest, and act to deliver electrical impulses to regulate heart rate. Implantable pacemakers are a well-established and generally safe medical treatment that nonetheless comes with risks.

“Pacemakers are one of the most important and widely used human implants, and they have saved millions of lives,” the paper’s co-corresponding author, Gengxi Lu, says. “But they are invasive, and they make direct contact with the beating heart. The dream for many years has been noninvasive heart stimulation with ultrasound.”

Ultrasound encompasses a range of acoustic waves that safely penetrates the body. Ultrasound waves reflect and resonate off structures in characteristic ways that allow technicians to resolve and image organs and tissues inside the body. Ultrasound can also be directed and focused to stimulate certain therapeutic effects, for instance in the brain, where scientists are exploring the use of ultrasound to treat Parkinson’s disease, Alzheimer’s, and other brain disorders.

Scientists have also found that ultrasound can benefit the heart. Previous studies in animals have shown that focused ultrasound can safely activate heart cells, though the effect has been inconsistent and weak.

Zhao and his colleagues looked to amplify ultrasound’s effects on the heart. In their new study, they applied sonogenetics, which is a relatively new approach that takes after optogenetics — a technique that involves genetically manipulating specific parts of a cell to respond to light. Similarly, sonogenetics aims to genetically engineer cells to respond to sound, including ultrasound.

In their work to develop an ultrasound pacemaker, the team first looked to increase heart cells’ sensitivity to ultrasound, through sonogenetics. In the lab, they used standard practices to derive heart cells from embryonic stem cells, and then delivered a genetic alteration to the cells that increased their sensitivity to ultrasound. Specifically, the manipulation produced ion channels that opened more readily in response to ultrasound.

“These channels can now ‘hear’ ultrasound better, and can open to let calcium in, which is what directly activates the cell and causes it to beat,” explains by the paper’s first author, Chen Gong.

Sticker health

In experiments with sonogenetically engineered heart cells, the researchers found that when they exposed the cells to ultrasound, the cells beat in sync with the waves, unlike cells that were not genetically manipulated.

In any clinical application of an ultrasound pacemaker, the team envisions that a patient could first receive a one-time injection, similar to a vaccine, that would act to genetically boost the sensitivity of cardiac cells to the pacemaker’s ultrasound waves. The injection would be a form of gene therapy — a treatment that is currently approved by the FDA to treat certain inherited conditions such as sickle cell disease and spinal muscular dystrophy.

“We think this step would be clinically translatable as a form of gene therapy that could enable noninvasive pacemakers,” Gong says.

The team then designed the core of the ultrasound pacemaker, in the form of a postage-stamp-sized sticker embedded with tiny ultrasound transducers. The sticky part of the device is made from a hydrogel material that Zhao’s group has refined over the years to adhere strongly to skin and many types of materials, while also allowing ultrasound waves to pass through without weakening. The transducers within the sticker can be tuned to generate ultrasound waves at specific frequencies.

In experiments with rats, the researchers first administered a sonogenetic, ultrasound-boosting solution through their tails. They then adhered a miniature version of the pacemaker to the rats’ chests. When they turned the stickers on, they observed that the ultrasound quickly regulated the animals’ hearts. Some individuals with slow heart rates were brought up to a normal rate, while others with irregular heartbeats were steadied, keeping in sync with ultrasound’s “ticks.”

“We can now use low-intensity ultrasound to open ion channels in cells to have very effective heart pacing,” Gong says. “We are now making these stickers into smaller form factors, and more integrated, so they are easier to wear, more stable, and more accurate over a longer term.”

“In this paper, we demonstrated noninvasive pacemaking. However, we think this concept could be useful beyond just the heart,” Zhao says. “We believe you could one day have stickers over different parts of the body that could do long-term imaging, monitoring, and closed-loop therapeutic stimulation.”

This work was supported, in part, by the National Institutes of Health, the National Science Foundation, the Department of Opthamology from Research to Prevent Blindness, and the U.S. Department of War.

A plan to preserve wetlands without stopping development

Peter Dizikes | MIT News — Tue, 02 Jun 2026 00:00:00 -0400

Balancing economic growth and environmental protection is not easy. Consider wetlands, which provide flood protection, aid water quality, and are linchpins of larger ecosystems. How can we best preserve wetlands while enhancing economic activity?

According to a new study, one solution involves supplanting traditional conservation mandates, which require replacing affected wetlands locally, with tradeable offsets. Through this system, a developer can build on a wetland by purchasing credits representing an equivalent environmental value created by improving a wetland somewhere else in the same watershed, away from concentrated development.

While this has largely been the approach of U.S. federal and state regulators since the mid-1990s, current regulations do not account for the flood protection benefits of wetlands. The new study finds a workable solution in an offset policy that also includes a locally varying tax on development, precisely to compensate for the increased flood risk it causes.

In the lower 48 states of the U.S., wetlands are heavily concentrated in California and Florida, two high-population states. Through a highly granular look at Florida’s wetlands from 1995 to 2020, with a new scholarly methodology that carefully weighs local factors, the scholars estimate that development of wetlands led to $2.4 billion in net economic gains. Their alternate policy would have preserved most of these gains while also preventing about $1.6 billion in flood damage.

“You’re retaining two-thirds of the private gains from trade,” says Daniel Aronoff PhD ’22, a research affiliate in MIT’s Department of Economics and co-author of a newly published paper summarizing the study’s findings. “And the flood damages shrink by an order of magnitude, so only you’re incurring a small fraction of the flood damage while collecting that amount in increased tax revenue, which can subsidize the cost of restoration after flood damage has occurred.”

This system is neither a simple conservation mandate nor a free ride for developers. The scholars say it would provide a better way of balancing wetlands preservation and economic gains, while lowering flood risk.

“You could do this,” Aronoff says. “It’s an implementable thing. You could build a policy out of this.”

The paper, “Conservation Priorities and Environmental Offsets: Markets for Florida Wetlands,” appears in the May issue of the American Economic Review. The authors are Aronoff, who is also a research associate at the Laboratory for Economic Analysis and Design at MIT and a research collaborator at the Digital Currency Initiative; and Will Rafey PhD ’20, an assistant professor of economics at the University of California at Los Angeles.

No net loss — but more risk

Federal wetlands policy in the U.S. has been governed since the 1970s by a “no net loss” objective, meaning that development must be accompanied by approved actions to offset any loss of wetlands functionality. State laws have often mirrored this federal approach. The current rules work on a watershed level, enabling public and private developers to offset the impact of developing a wetland by purchasing offset credits from a “wetland mitigation bank” in the same watershed.

The researchers developed their study as an ambitious, data-rich project. They obtained comprehensive data on environmental offset credits issued, and transfers to developers from state and regional regulators; a record of offset prices from a private broker as well as state and county purchase records; maps detailing wetlands development and private property ownership; and Federal Emergency Management Agency (FEMA) data on flood risk policies and claims.

The scholars then built a detailed database of development from every wetland bank permit issued in Florida that included enhancements, land acquisition, estimated costs, and offset credit release schedules, as well as records of actual releases and sales over time. They used these data to build a dynamic model of the wetland offset market, from which they obtained their estimates of economic gains and flood risk costs.

Whereas other work has applied national data to wetlands analysis, this more granular approach allowed the scholars to conduct a locally focused examination of economic activity, floods, and policy specifically applying to Florida.

“The functional form that has been used to estimate the relationship between wetlands and flood risk across all America is not compatible with data on wetlands and flooding in Florida,” Aronoff says.

The study also underscores an important distinction in the kinds of offset policies that have previously been deployed. The first iteration of offset policy required a developer to restore wetlands adjacent to any wetlands area that is newly developed. A second iteration, the one still in use, allows developers to purchase offset credits — which might apply to wetlands that are not adjacent to the development in question. The latter carries with it greater risk of flood damage to developed property, as an equivalent amount of restored wetlands in a rural area will not serve as a flood buffer for as many structures.

The proposed policy solution would levy a tax — either on offset sellers or buyers — that would equal the estimated increase in flood risk created by the development.

“Going from the first policy iteration to the second iteration could have created a lot of value, because you have development taking place with wetlands created in the lowest-cost way,” Aronoff says. “But that gave rise to an externality: the flood risk. Because you’re creating flood risk by developing in urban areas with lots of buildings, while creating wetlands in rural areas without buildings around.”

Tuning the policy

Ultimately, that is why the empirical analysis developed by the economists shows a more optimal path using so-called Pigouvian taxes, named after 20th-century economist Arthur Pigou. These taxes add a levy when people create negative circumstances for society at large. Taxes to inhibit pollution, for instance, are Pigouvian. The modeling in the current study indicates the same concept would work effectively for wetlands policy.

“Economics is about tradeoffs,” Aronoff says. “And this is a tradeoff. Flood risk is expensive — that’s a cost. But development creates value because it is only profitable to the extent that the end user desires it.”

Ultimately, the scholars think, implementing systems that balance factors will work better in the long run than many kinds of prohibitions on economic activity — or than allowing unrestricted activity without weighing the public good.

“If you choose an absolute, you’re choosing one over the other in all instances,” Aronoff says. “And what is at the core of the outlook of an economist is to assume there’s a tradeoff, and the question is how do you negotiate that tradeoff in an optimal way. That’s what we are trying to get at here.”

The research was supported by the National Science Foundation and the George and Obie Schultz Fund.

New propulsion system could make tiny satellites both fast and fuel-efficient

Jennifer Chu | MIT News — Mon, 01 Jun 2026 17:15:00 -0400

MIT engineers are testing a new propulsion system that combines the power and speed of conventional chemical thrusters with the precision and fuel-efficiency of electrical thrusters.

The system could enable the design of nimbler, more flexible small satellites, which could perform both fast, powerful maneuvers and slower, precise adjustments, depending on the mission and moment at hand.

The key to the new system is a special propellant that can power both chemical and electrical thrusters, which traditionally have required separate, bulky fuel sources.

“If you can have chemical and electrical propulsion in one small package, it’s the best of both worlds,” says Amelia Bruno, a former postdoc in MIT’s Department of Aeronautics and Astronautics (AeroAstro). “This opens the door for small satellites to do even more science, more observations, and more interesting missions, all on a smaller and cheaper platform.”

Bruno is the lead author of a study appearing this week in the Journal of Propulsion and Power showing that a type of “green monopropellant” originally developed by the U.S. Air Force for use in chemical propulsion in space can also effectively power tiny “electrospray” thrusters. Electrospray thrusters are dime-sized rockets that use electric fields to charge up a liquid propellant’s particles, which are then shot into space as a thrust-generating spray.

Electrospray thrusters are extremely fuel-efficient and can perform slow and precise maneuvers, such as pushing a small spacecraft bit by bit through a long, interplanetary journey. Chemical thrusters, in contrast, require a large fuel supply to perform short and fast bursts, for instance to quickly ascend and descend, or speed up and slow down.

Now that the MIT group has found a propellant that can fuel both chemical and electrospray thrusters, they see big potential for small spacecraft. The team is working with NASA to launch the Green Propulsion Dual Mode mission — a briefcase-sized CubeSat that will carry a chemical thruster and four electrospray thrusters, all fueled by a single propellant tank. The mission will be the first to test such a two-in-one propulsion system for small spacecraft. If it is successful, Bruno says the mission could pave the way for small satellites to explore beyond Earth’s orbit.

“We could send CubeSats to Mars, or the asteroid belt, where they could make the journey slowly, using electrospray thrusters,” says study co-author Paulo Lozano, the Miguel Alemán Velasco Professor of Aeronautics and Astronautics at MIT. “You could then use your chemical thrusters to quickly move to look at interesting features. You could have a lot more flexibility to do a lot more things.”

The study’s co-authors also include Matthew Corrado SM ’22, PhD ’26.

A sea of ions

Lozano’s group at MIT designs, fabricates, and tests electrospray thrusters for use in satellites that range from the size of a lunchbox to a small carry-on suitcase. Compared to conventional satellites, these microsatellites are significantly smaller and cheaper to launch into space.

But smaller spacecraft require smaller everything else, including propulsion systems. In that respect, electrospray thrusters are a good fit. The thrusters Lozano develops are about the size of a thumbnail. Each thruster sits atop a small reservoir of ionic liquid propellant. When the reservoir is connected to a battery, the battery supplies some amount of voltage that electrically charges a corresponding amount of ions in the liquid. The charged particles are then channeled out of the reservoir, through the thruster’s tips and into space as a thrust-inducing spray.

Over the past decade, Lozano has tested many thruster designs, under varying conditions, and with various types of ionic liquid propellant — a fuel that is essentially made from salts that can remain in liquid form.

“Ionic liquids are very stable and can even remain a liquid in space, which not a lot of materials can do,” Bruno says. “And it’s basically a sea of ions, which is why we base our technology around it, so we can pull those ions out into an electrospray.”

Bruno and Lozano have collaborated with the U.S. Air Force, which synthesized a new kind of ionic liquid propellant — the Advanced SpaceCraft Energetic Non-Toxic propellant (ASCENT) — which was being tested in chemical thrusters. Chemical thrusters are high-force propulsion systems typically associated with launching rockets and performing hard and fast maneuvers once in space. ASCENT was designed as a “green,” less toxic alternative to hydrazine, which has been the traditional fuel source for chemical propulsion and is extremely hazardous to handle.

“ASCENT happens to be an ionic liquid mixture,” Bruno says. “And we said, hey, that’s the stuff we typically use. Theoretically, this should work. Let’s go figure out how.”

Spray and spin

In their new study, Bruno, Lozano, and Corrado tested the performance of electrospray thrusters that they fueled with ASCENT. Each thruster they used was attached to a small cube-shaped reservoir about the size of a Lego brick. They filled each reservoir with 1 gram of ASCENT, a liquid that has a viscosity similar to baby oil. They then attached a thruster to opposite sides of a CubeSat, which they set on a MagLev stand — a custom testbed that is designed to magnetically levitate a sample or device. The MagLev in Lozano’s lab is installed inside a large vacuum chamber, which the researchers can tune to mimic the conditions in space.

Over multiple experiments, the team remotely applied varying levels of voltage to activate the thrusters, which in turn produced a spray that spun the CubeSat around, like a floating, spinning top. The researchers measured the amount of thrust produced with each trial, and calculated ASCENT’s fuel efficiency as they ran the thrusters continuously over periods lasting up to 100 hours.

In the end, they found that ASCENT was able to successfully fuel each electrospray thruster. What’s more, the propellant, which was originally intended for chemical propulsion, was just as efficient as other, conventional ionic liquids at propelling electric thrusters.

“Compared to our normal electrospray propellants, ASCENT can provide similar performance in terms of thrust,” Bruno says. “Now that we know our thrusters work with ASCENT, we can start thinking of all the ways we can make them even better.”

Now that ASCENT has been proven to work in both chemical and electrical propulsion, she and Lozano say that a single tank of the fuel can be used to power both types of thrusters, all in a compact, two-in-one system that could fit within a small CubeSat. The team will test the idea with NASA’s Green Propulsion Dual Mode mission, which is scheduled to launch in November.

“This will be the first time that a satellite will have a shared propellant tank,” says Lozano, who notes that in addition to long, exploratory interplanetary missions, small satellites equipped with both chemical and electrical propulsion could also be useful for missions closer to Earth, such as for weather and climate observations.

“Say there’s a storm coming, and you’d want to deploy your constellation of small satellites to observe over one location,” he says. “You could choose to send them quickly or slowly depending on the nature of the observation. And the only way to do that is if you have two propulsion systems, which is now possible.”

This research is supported, in part, by NASA.

Enzymes that assemble into droplets can speed up cellular reactions

Anne Trafton | MIT News — Mon, 01 Jun 2026 11:00:00 -0400

Within the past decade, biologists have discovered that one strategy cells use to keep their contents organized is a phenomenon known as phase separation.

Similar to the way oil forms droplets that float in a vinegar solution, proteins inside cells can phase separate to form highly concentrated droplets that keep them organized within the cell. In a new study, MIT researchers have now shown that this droplet formation is critical for controlling the function of a class of enzymes called kinases.

The researchers found that condensing into droplets optimizes the biochemical conditions needed for kinases to catalyze reactions, allowing them to more rapidly activate cell signaling pathways. In some cases, droplet formation can even change which reactions the kinases perform.

“Many biological molecules have this propensity to spontaneously separate. We were really interested in asking, if we have these kinases forming droplets, what is the consequence of that in the context of signaling?” says Lindsay Case, an assistant professor of biology at MIT and the senior author of the study.

Learning more about how these droplets form could help researchers design drugs that target kinases, some of which can be overactive in cancer cells.

“Understanding the chemistry of these compartments, and what molecules go into them and what molecules don’t go into them, could help us design drugs that better localize to their target of interest,” Case says.

Nicholas Lea, an MIT graduate student, is the lead author of the paper, which appears today in Cell Reports.

Forming droplets

Since her days as a graduate student, Case has been studying how the physical organization of molecules inside cells affects their function. As a postdoc, she began studying how phase separation might affect a signaling pathway that allows cells to sense when they’re attached to their environment, so they can respond appropriately.

Some of the proteins in this pathway are kinases, which activate other proteins by adding phosphate groups to them. Kinases can also activate themselves through a process called autophosphorylation.

“Inside of the cell, you have these kinase molecules that are responsible for carrying a signal through the cell, and we know that the organization of these molecules changes. When the information is present, they’re organized in a different way than when the information is not present,” Case says. “We think that having the right molecules in the right place is incredibly important for the right biochemistry to occur.”

Phase separation is one of the methods that cells appear to use for this organization. The most familiar example of phase separation can be seen in a salad dressing, where oil forms droplets to minimize contact with water-based vinegar. Proteins can phase separate when they are highly concentrated, leading them to self-assemble into dense droplets floating in the cell’s cytoplasm.

Case hypothesized that this phase separation, which brings kinases together at a high density, might help cells to boost the enzymes’ activity because they are more likely to bump into and phosphorylate each other.

In this study, Case and Lea set out to test that hypothesis, focusing on an enzyme called focal adhesion kinase (FAK). This kinase, which becomes activated when cells attach to their surrounding environment, activates pro-growth and pro-survival signals. In cancer cells, this signaling pathway can go awry, allowing cells to proliferate even when they detach from their original locations.

Scientists already knew that when cells are properly attached to their environment, that adhesion signal causes FAK to accumulate at the cell membrane. In the new study, the MIT team mimicked that effect by overexpressing FAK in cells. These cells were floating freely in a solution, not attached to any surface. Even so, the high concentration of FAK caused the kinase to phase separate into droplets, which turned on the pro-growth signal.

“It was surprising that just by condensing this protein into a droplet, you can actually turn on a signaling pathway that should be turned off,” Case says. “If FAK concentration is too high, you’re always getting these droplets and you’re always signaling, regardless of what the receptors that are supposed to be controlling this are doing.”

The findings suggest that in cancer cells, overexpression of FAK may lead to phase separation, which then helps to drive cancer progression and metastasis.

“It may be that for some kinases, you’re not supposed to form these droplets in the cytoplasm because it leads to this always-on signal, and then the cells no longer listen to the information coming from the environment,” Case says.

Interfering with FAK’s ability to form droplets could offer a new strategy for cancer drug development, she says.

Controlling reactions

The researchers also studied two other kinases, Mst2 and Abl. They found that these enzymes could also phase separate at high concentrations, and that this increased their activity. While phase separation of FAK in the cytoplasm may occur only in cancerous cells, for Mst2, it appears to be a strategy that healthy cells use to control a signaling pathway called Hippo, which promotes cell growth and survival.

Additionally, for both Mst2 and Abl, the researchers discovered that phase separation can lead the enzymes to phosphorylate additional targets, which may lead them to activate different signaling pathways.

“It’s not just that you’re getting faster phosphorylation, but in those cases, the patterns of what is actually getting phosphorylated were very different inside of the droplet compared to what might be happening in a non-droplet context,” Case says. “The kinase is able to phosphorylate amino acid residues beyond the set of canonical sites that have been described before.”

The researchers also found that when these droplets form, they attract high concentrations of ATP, the molecule that kinases use as a source of phosphate. This occurs because kinases tend to contain floppy sections containing many positively charged amino acids, which attract negatively charged ATP.

Using a machine-learning model, the researchers predicted that about 45 percent of the 500 kinases found in human cells would have the ability to form droplets like those seen in this study. Those kinases were also more likely to be highly positively charged, which could help them to recruit ATP into the droplets.

In future work, Case hopes to explore the possibility of designing drugs that could mimic ATP’s ability to be attracted into droplets within a cell, which could help reduce negative side effects of the drugs.

“By localizing drugs to the compartment where your target localizes, that could reduce off-target effects by concentrating the drug with the target of interest and reducing interactions with other molecules,” Case says.

The research was funded by a Searle Scholars Program Award, the U.S. Air Force Office of Scientific Research, the National Institutes of Health, the Royal G. and Mae H. Westaway Family Memorial Fund, and a David H. Koch Graduate Fellowship.

New laboratory at MIT aims to advance quantum research for the nation

Zach Winn | Abby Abazorius | MIT News — Thu, 28 May 2026 17:20:00 -0400

On May 28, MIT President Sally Kornbluth and Massachusetts Governor Maura Healey announced plans for a new laboratory to accelerate the development of next-generation quantum technologies that will enable Massachusetts to remain a national hub for quantum innovation.

Speaking at the Samberg Conference Center on campus, the leaders introduced the Quantum Systems Laboratory (QSL) at MIT, a shared-use facility that will catalyze quantum development in the region and help keep America at the forefront of a technology seen as critical for a range of industries.

“Quantum technologies have the potential to drive transformative change in fields from computing, security, and navigation to health sciences, defense technologies, and space exploration,” Kornbluth said. “Greater Boston has the greatest concentration of quantum talent of anywhere in the world, so it has been clear to us for some time that if we could magnify all of that talent with the right facilities — a shared quantum toolbox — we could establish Massachusetts as a national hub for quantum innovation and help catalyze the next generation of quantum technologies.”

The Quantum Systems Laboratory will join a state-of-the-art quantum computer with the components needed to make it a scalable, practical technology for solving complex, real-world problems. Such components include peripheral hardware such as sensors and quantum interconnects, which are physical channels that transfer quantum information. Located at MIT’s Building 39, the facilities will be open to researchers both from and beyond MIT.

Thanks to a $25 million investment from the state, announced today, which will match a portion of the federal funding for quantum research already underway at MIT, the Institute is now in a position to move forward as early as this summer with construction on the QSL facility. The Commonwealth’s investment adds to MIT’s own financial commitment, as well as generous philanthropic support from Thomas Tull.

“This is good news for MIT, good news for Massachusetts, and frankly, good news for the world that we’re working together to make this happen,” Healey said. “The return on investment is clear: We know the Quantum Systems Laboratory will be a first-of-its-kind center for the shared study and development of quantum science and technology. It’s going to unleash the great power of scientists and innovators from around the state and across the world, and also be a place for collaboration, both for academic and commercial ventures. It will offer incredible opportunities for both scientific progress and economic growth. It’s a testament to MIT’s unrelenting, unyielding belief in the power of openness and collaboration to advance science.”

The new lab will be the physical home for the MIT Quantum Initiative (or QMIT) announced by President Kornbluth in December. It also complements advanced facilities already used for quantum research at MIT, such as MIT.nano and MIT Lincoln Laboratory’s SQUILL foundry, both of which share the mission of democratizing access to world-class facilities. SQUILL and MIT.nano have already made a major impact on the quantum industry through research, startups, and new standards for creating and transmitting quantum information.

“I want to emphasize that just as MIT.nano is a facility for all, there will be many people from beyond MIT that come to use this equipment” at QSL, Kornbluth said. “This is a hub to make Massachusetts the center of the world for quantum. These resources are rare enough that we have to make sure they are available to our colleagues at the University of Massachusetts, Harvard, and beyond. Our plan is to mobilize all the talent in the area through this facility.”

Leading in quantum innovation is important for the prosperity and security of the country, but quantum research requires meticulously controlled environments. The new facilities will give scientists access to the cutting-edge quantum hardware and specialized experimental capabilities needed to achieve the full transformative potential of quantum science and engineering.

The new laboratory’s underlying mission is to return broad scientific, workforce, and economic benefit to the public.

For example, quantum technologies provide significant opportunities in the fields of life sciences and defense technologies, which are $50-billion contributors to the local economy, with dozens of startups working in the area. The new lab is designed to create new job opportunities in the form of academic research, startups, and more. Construction on the QSL facility alone is anticipated to create over 150 full-time, on-site jobs, plus another 75 to 100 jobs across the Commonwealth in supply chain and professional services supporting the project.

Startups from MIT are also a key driver of the region’s entrepreneurial ecosystem; in 2015, Sloan Professors Edward Roberts and Fiona Murray published a report detailing how the Institute’s alumni entrepreneurs have created more than 30,000 active companies, employing 4.6 million people and generating annual global revenues of $1.9 trillion, a figure greater than the gross domestic product (GDP) of the world’s 10th-largest economy, as of 2014. The QSL facility will provide the necessary equipment and facilities for startups working on quantum technologies, thereby strengthening the region’s innovation economy.

MIT researchers develop a low-cost technique to get lithium out of rocks

Zach Winn | MIT News — Thu, 28 May 2026 14:00:00 -0400

Demand for lithium has surged in recent years as lithium-ion batteries power increasingly more of our world. And yet, even as places like the U.S., Europe, and Australia have abundant lithium resources within their borders, China dominates global lithium refining. The biggest hurdle to tapping into the U.S. and Australia’s lithium is getting it out of hard rock minerals in a form that is useful.

Extracting lithium from hard rock today is an energy- and waste-intensive process that is often far more expensive than getting lithium from brine water, which also has major environmental drawbacks. Currently, lithium hard rock extraction involves baking the rock at over 1,000 Celsius and chemically leaching it to extract lithium. The rest of the rock is discarded.

Now, a team of researchers from MIT and elsewhere has developed a low-temperature process for extracting battery-grade lithium from the most common type of lithium-bearing mineral. The process uses a liquid reagent to dissolve the rock into the useful forms of its constituent parts: not just battery-ready lithium salts, but also smelter-grade alumina and cement-ready silica. After the minerals are extracted, the solvent and reagent can be recovered and used again so waste levels approach zero.

The researchers estimate the closed-loop process is half the cost of traditional lithium hard rock extraction and could make it cost-competitive with extracting lithium from brine water.

A paper describing the process was published today in Science. The researchers have already begun commercializing the technology through an MIT spinout, Rock Zero.

“By 2040, we need to quadruple production of lithium globally, which amounts to hundreds of new lithium producing assets,” says author Camden Hunt, a former project manager in MIT’s Center for Electrification and Decarbonization of Industry. “Hard rock is abundant; you can find it everywhere. But most hard rock refining is done in China. Our central thesis is if you can find an easier way to crack the rock, get lithium out, and make battery-grade lithium salts, you can change the lithium market. It aligns with the recent push to onshore production of critical minerals in the U.S.”

Joining Hunt on the paper are former MIT postdoc Benjamin Mowbray; PhD candidate Kalyn Fuelling; MIT undergraduate Jacqueline Prawira; Khashayar Jafari, a former senior research scientist at the MIT green cement spinout Sublime Systems; and Yet-Ming Chiang, MIT’s Kyocera Professor of Materials Science and Engineering.

From bathrooms to batteries

The research has its roots in a bathroom renovation. About 25 years ago, as Chiang made a trip to a hardware store to look for something that would turn clear glass blocks translucent, he stumbled on a glass etching cream that works by “eating away” at the surface of the glass. The active ingredient turned out to be ammonium fluoride.

More recently, as Chiang was brainstorming ways to chemically break apart the most abundant lithium-bearing mineral, spodumene, he thought back to that etching cream. Spodumene, like glass, consists mostly of silica. Conventional chemistry-based methods for extracting metals from ores preferentially dissolve more reactive elements and leave behind a silica-enriched residue because of the strength of silicon-oxygen bonds. By designing their process to use a mixture of water and ammonium fluoride, the researchers are able to dissolve silica first, reversing the process.

The researchers showed they could dissolve spodumene rock at room temperature, which represented a breakthrough over traditional processes requiring extreme heat. But it was still only the first step to a closed-loop system that produced useful materials.

“Dissolving silica is the hard part in mining,” Mowbray says. “The next question was how do we apply it to impactful mineral processing problems?”

The mineral spodumene is mainly made up of three components: lithium, aluminum, and silica. Mowbray and Hunt, who both have their PhDs in chemistry, began exploring ways to refine those components separately after they were broken apart in the ammonium fluoride solution.

First, the researchers isolated lithium fluoride, a useful input for common electrolyte materials used in batteries. Chiang, who has founded several battery companies over his multi-decade career at MIT, next asked the research team if they could isolate lithium hydroxide and lithium carbonate, two lithium salts useful for making battery cathodes. The researchers went back to the lab and found they could make both by developing new processes, some of which involved adding carbon dioxide or sodium carbonate. Chiang tasked the research team with a similar challenge for the aluminum part of the rock, which was isolated using a high-temperature separation technique, and then silica, which was isolated by precipitation.

“First our goal was to produce these products, then there were additional steps of characterizing their purity and properties and making sure our products met the specifications for target markets,” Mowbray explains. “For the lithium salts, we identified the purity specifications for battery-grade lithium carbonate, the most widely used lithium salt. For the silica, we wanted it to be used as a cement additive, so we did cement reactivity tests and eventually created cubes of cement from it for strength testing using industrial methods. For aluminum, we targeted smelter-grade aluminum. If any product didn’t meet the target specs, you’d end up with a waste stream.”

The researchers then developed a process to reuse the ammonium fluoride and water that starts the reaction.

“We’re able to dissolve the rock with the spodumene in it, and that liberates all the elements, including the aluminum and lithium,” Chiang says. “The silica is in the solution, but on the way to making ammonium fluoride, ammonia gas also comes off. If that ammonia gas is then reapplied, it precipitates the silica again. That sequence gives us back the starting ammonium fluoride. That’s why it’s a circular process.”

The researchers successfully processed 17 different spodumene rock sources, showing its widespread applicability using rocks around the world.

“You’ve heard of nose-to-tail eating?” Chiang says. “We refer to this as nose-to-tail mining. Our researchers came to MIT to look for impactful problems to work on in sustainability. With their skill sets, it was just a matter of setting them loose on this problem. We went through all these steps, and for each one, I’d just say, ‘Can you do this next step?’ And a week or two later they’d say, ‘Okay, we’ve shown we can do that.’ That’s how this entire process got built.”

Scaling the process

Chiang further challenged his research team to evaluate the commercial feasibility of their new system.

“Once we had these core operations worked out, Yet encouraged us to do some math,” Mowbray explains. “Is there enough spodumene in the world to supply 100 terrawatt-hours of battery production? The follow up was: If you supply all the world’s batteries with this process, what are the volumes of the co-products? Do they match global commodity markets? Then we started looking at the cost of the reagents, the cost of the energy, equipment. We started gaining conviction that this could have a big impact.”

The work has special significance for Mowbray, who grew up in a historic mining town in rural British Columbia.

The researchers worked with MIT’s Technology Licensing Office to spin out their company, Rock Zero, which is now located at The Engine and scaling up the system.

“We believe this approach is the lowest-energy, lowest-cost way of getting lithium not only out of hard rock, but period,” Chiang says. “That’s what’s motivating us to scale this. It will enable the energy transition through batteries that use lithium. This was one of the goals of The Climate Project at MIT — to work on projects that, within a short number of years, could transition from the lab to commercialization and impact.”

The work was supported, in part, by the Department of Energy Advanced Research Projects Agency-Energy (ARPA-E), the MIT Climate Grant Challenges program, and the National Science Foundation. The work made use of MIT.nano facilities.

A new sensor could enable earlier detection of bladder cancer

Anne Trafton | MIT News — Thu, 28 May 2026 12:00:00 -0400

Every year, about 85,000 Americans are diagnosed with bladder cancer. While treatment is often successful, bladder cancer has one of the highest rates of recurrence of any cancer: Following treatment, about 50 percent of patients develop tumors again within the next five years. This makes it one of the most expensive cancers for society to treat.

MIT researchers have now developed a new way to regularly monitor those patients, which could enable regrowing tumors to be detected much earlier. Using a catheter coated with specialized nanosensors, the team showed that they could detect very low levels of a protein produced by bladder cancer cells and image their location in tissue.

The researchers calculate that this sensing approach is nearly 50,000 times more sensitive than urinalysis, an approach that has been used to monitor bladder cancer in patients. In an animal study, they showed that fluorescent signals produced by the sensors can be used to pinpoint the location of the tumor within the lining of the bladder, providing a chemical image.

“It’s like a camera for molecules instead of light,” says Michael Strano, the Carbon P. Dubbs Professor of Chemical Engineering at MIT. “If you have a billion nanosensors in an array, you can use them to make a chemical image that helps you locate their source.”

Strano is the senior author of the study, which appears today in the journal Nature Nanotechnology. Wonjun Yim, a Schmidt Science postdoc, and Hohyung Kang, an MIT postdoc, are the lead authors of the paper. Other authors include MIT graduate student Marco Machado, undergraduate student Maeve McGinnis, and postdoc Byungha Kang.

“Chemical images”

The new detection approach is based on carbon nanotubes — hollow, nanometer-thick cylinders made of carbon that naturally fluoresce when exposed to laser light. Over the past 10 years, Strano’s lab has shown that these nanotubes can be customized to sense different molecules by coating them with “synthetic antibodies” — polymers that can be designed to interact with a specific target.

When the target analytes are present, their interaction with the synthetic antibodies causes the carbon nanotubes to shift the wavelength or change the fluorescent intensity that they produce. Strano’s lab has previously developed about two dozen different sensors that can detect different targets, including hydrogen peroxide, riboflavin, and viral proteins.

For the new study, the researchers designed a sensor that could detect a protein known as nuclear matrix protein 22 (NMP-22), which is already FDA-approved for use as a biomarker for bladder cancer. NMP-22 can be detected in urine samples, but it is often significantly diluted, degraded, and cleared after secretion. This means that tumors can only be detected once they have reached more advanced stages.

To enable earlier detection, the MIT team sought a way to deploy their sensors inside the bladder, where they could detect NMP-22 near the tumor at locally elevated concentrations. The device they designed consists of a urinary catheter coated with nanotubes that can sense NMP-22. The catheter also contains a tiny device known as a ball lens, located within the tip of the catheter.

This lens rotates 360 degrees, emitting laser light and then absorbing the fluorescent light emitted by the nanosensors. By analyzing the color and location of these fluorescent signals, the researchers can map the location of any biomarker that is detected.

These chemical images can reveal not only whether the biomarker is present, but also the location of the cancerous cells.

“If you are scanning over a region of tissue, you would like to know not just that there is a signal indicating that a tumor is there, but also its location so that you can treat it or perform a biopsy,” Strano says. “Before an early-stage tumor breaks through the urothelium so that it’s visible, it’s under the surface but still emitting chemical signals that can be imaged. When a chemical hits the catheter, we don’t just detect its presence, but we collect a map that pinpoints its location.”

Tests in animal bladders showed that this type of detection can be 180 times more sensitive than performing a conventional urinalysis because it detects biomarkers directly where they are produced in the bladder, rather than measuring them later in dilute fluids such as urine, where their concentration is much lower. This high degree of sensitivity would allow the sensors to detect signals from a tumor as small as 16 square millimeters, the researchers say.

Earlier detection

Researchers in Strano’s lab are now working on designing a more compact version of their prototype imaging system, so that it could be used more easily at a doctor’s office. They also hope to incorporate their sensors into a type of catheter known as a cystoscope, which has a camera attached and is used to visualize tumors in the lining of the bladder.

Currently, patients who have been treated for bladder cancer undergo cystoscopy annually, or in some cases even more often, to monitor for cancer recurrence. The new MIT diagnostics should be able to detect recurring tumors earlier than cystoscopy, making them easier to treat and cutting down on the costs of treatment and monitoring, the researchers say.

“What we’re looking for is something that could be faster and more effective. It could be used right in a doctor’s office, and it could make that screening more efficient and less invasive, with much lower cost. The goal is to be able to detect potential tumors much earlier,” Strano says.

“This paper is exciting because it shows how diagnostics can be more effective when the sensor is brought to the individual,” says Daniel Heller, a professor of physiology and pharmacology at Weill Cornell Medicine, who was not involved in the research. “Strano and colleagues demonstrated that a carbon nanotube-based nanosensor technology can be used to monitor a cancer right where it is, improving the speed of cancer detection, and potentially enabling the improvement of cancer treatment.”

This approach could also be integrated with endoscopy to detect other types of cancer or other diseases, such as cardiovascular or gastrointestinal diseases, by swapping out the nanosensors attached to the catheter.

“The beauty of polymer chemistry is that if we understand the molecular structures of target biomarkers and the design principles of binding sites, we can develop new sensors tailored to different diseases,” Yim says. “You can imagine if these sensors were integrated onto the catheter, they could reveal invisible biomarkers that current endoscopic procedures miss, opening the door to detecting many other diseases in the future.”

The research was funded by the Bridge Project of the Koch Institute and Dana-Farber/Harvard Cancer Center, a Schmidt Science Fellowship, the MIT UROP Program, Mathworks Inc., and a National Science Foundation Graduate Research Fellowship.

Media Advisory: MIT to establish regional quantum hub

Thu, 28 May 2026 12:00:00 -0400

MIT and the Commonwealth of Massachusetts announced plans to establish the Quantum Systems Laboratory (QSL) at MIT, which will be open to researchers across the region.
With the new funding from the state, which will match federal funding for quantum research already underway at MIT, the Institute aims to begin construction on the QSL facility this summer.
The QSL will host specialized facilities that will enable Massachusetts scientists to undertake impactful work applying quantum research across practical domains, including life sciences and national defense.

Quantum technologies promise transformative changes in fields from computing, security, and navigation to health sciences, defense technologies, and space exploration. But how do we ensure Massachusetts stays on the leading edge of our nation’s coming quantum leap? Doing so is vital to the prosperity and security of our Commonwealth and country, serving to protect and advance America’s technological leadership in a world that has been upended by geopolitical rivalries.

On Thursday, May 28, Governor Maura Healey joined President Sally Kornbluth at MIT to announce a new effort aimed at establishing Massachusetts as a national hub for quantum innovation and catalyzing next generation quantum technologies. MIT and the Commonwealth of Massachusetts announced plans to establish the Quantum Systems Laboratory (QSL) at MIT, a new shared-use facility that will serve as a quantum toolbox for the region, aimed at accelerating quantum research, innovation, and growth in this critical field.

The QSL seeks to be the first facility in the world to bring together state‑of‑the‑art quantum computers with quantum sensors and peripherals, joined by quantum interconnects (physical channels that transfer quantum information). The facility will provide researchers from MIT and other institutions hands‑on access to significant quantum hardware and specialized experimental capabilities that are necessary to achieve the full transformative potential of quantum science and engineering.

Thanks to a $25 million investment from the state, which will match a portion of the federal funding for quantum research already underway at MIT, the Institute is now in a position to move forward as early as this summer with construction on the QSL facility, positioning the region to dominate the next generation of quantum research, according to Institute officials. The Commonwealth’s investment adds to MIT’s own financial commitment, as well as generous philanthropic support from Thomas Tull.

“Greater Boston has the greatest concentration of quantum talent anywhere in the world, working on a range of potential applications. Through the new Quantum Systems Laboratory, we will help position Massachusetts to lead the next era of quantum technologies,” says Kornbluth. “This facility will serve those at the edges of our wildest imaginations in physics and quantum computing, yes. But it will also equip the talent in our region -- and ultimately, our nation -- to push our knowledge to new limits, and new innovations.”

The QSL will be located at Building 39 on the MIT campus and will serve as a multi-disciplinary quantum hub with modern experimental infrastructure. Because quantum research involves the creation and study of coherent phenomena in systems that are isolated from the rest of the universe, it must take place in a highly controlled environment. Work is already underway in Building 39, with significant investments by MIT, to upgrade the physical infrastructure for these unique demands. The state’s support will supercharge this work and allow for the transformation of the lab into a hub for scientists across the region working on next-generation quantum technologies, startup applications, defense and health tech, and more.

“Our region has unparalleled strengths in science-intensive innovations and tough tech breakthroughs that combine engineering, science, and computing,” notes Anantha Chandrakasan, MIT’s provost. “With the new Quantum Systems Laboratory, we aim to arm Massachusetts with the compute power and integrated platforms needed to lead the coming era of quantum technologies.”

By the numbers

The QSL will host specialized facilities that will enable Massachusetts scientists to undertake impactful work applying quantum research across practical domains. As a shared-use facility, the QSL is being developed with the underlying mission of returning broad scientific, workforce, and economic benefit to the public.

For example, quantum technologies provide significant opportunities in the fields of life sciences and defense technologies, which are $50 billion contributors to the Massachusetts economy, with dozens of startups working in the area. During a time of increased economic anxiety and labor market concerns, investing in foundational quantum facilities will infuse our region with new job opportunities, in academic research institutions, startups and more. Construction on the QSL facility alone is anticipated to create over 150 full-time, on-site construction jobs, plus another 75 to 100 jobs across the Commonwealth in supply chain and professional services supporting the project.

Startups from MIT are also a key driver of the state’s entrepreneurial ecosystem; in 2015, Sloan Professors Edward Roberts and Fiona Murray published a report detailing how the Institute’s alumni entrepreneurs have created more than 30,000 active companies, employing 4.6 million people, and generating annual global revenues of $1.9 trillion, a figure greater than the gross domestic product (GDP) of the world’s 10th-largest economy, as of 2014. The QSL facility will provide the necessary equipment and facilities for startups working on quantum technologies, thereby strengthening the region’s innovation economy.

“The new QSL will introduce modern experimental infrastructure to quantum research at MIT and beyond, allowing us to scale experiments and expand into critical domains in disciplines such as biology and chemistry, where we see enormous innovative potential,” explains Ian Waitz, MIT’s vice president for research. “As the new physical home of the MIT Quantum Initiative (or QMIT), the QSL will serve not only as an on-campus incubator, but more broadly, a regional hub to catalyze quantum innovation, growth, and investment in this critical R&D sector for the Commonwealth.”

One floor of the facility will allow for development of radio-frequency (RF) electronics for controlling and interfacing with quantum systems. The QSL will also support researchers in the creation of customized quantum experiments with advanced high-frequency packages, which are required to protect quantum data in real-world applications. The facility will also develop the associated THz electronics needed by advanced quantum systems.

A history of future-focused plays

Nearly a decade ago, MIT made a similarly big bet on nanotechnology, developing MIT.nano — a state-of-the-art, shared-use facility with more than 200 tools and instruments that support nanoscale discovery and innovation through imaging, fabrication, characterization, and prototyping. Set in the heart of campus in the Lisa T. Su Building, MIT.nano is home to a thriving research community, an industry consortium, and a startup accelerator. More than a fifth of the 1,500 users of MIT.nano come from outside of MIT, and half of the companies in its START.nano accelerator have had non-MIT founders.

The QSL will also complement the capabilities of MIT Lincoln Laboratory’s SQUILL Foundry, a quantum fabrication hub for superconducting qubit systems that serves researchers across Massachusetts and the nation free of charge.

Brighter MRI signals

Jennifer Michalowski | McGovern Institute for Brain Research — Wed, 27 May 2026 16:40:00 -0400

When doctors and scientists want to see inside a body, magnetic resonance imaging (MRI) is a powerful tool. MRI can noninvasively capture detailed images of the body’s muscles, organs, and bones. It can monitor blood flow to generate a map of brain activity. And with new sensors developed by bioengineers at MIT, MRI can track the kinds of molecules that make our brains and bodies work.

In the May 13 issue of the journal Nature Biomedical Engineering, a team led by Alan Jasanoff, the Eugene McDermott Professor in the Brain Sciences and Human Behavior at MIT, reports on their new sensors, which can brighten or dim MRI signals in response to specific molecular targets. The probes are designed to amplify the effect that each target molecule has on MRI signal, dramatically improving sensitivity over previous small-molecule sensors. Jasanoff, who is also an associate investigator at the McGovern Institute for Brain Research, says the approach his team used should enable the development of MRI sensors that detect neurotransmitters and other important molecules in the brain.

“We want to be able to measure distinct chemical signals like neurotransmitters, neuropeptides, and metabolites as they fluctuate across the whole brain,” Jasanoff says. “These chemicals are important ingredients in neural computations, and we want to use the types of probes that we developed to detect these signals dynamically.”

Jasanoff explains that researchers have struggled to use MRI to sensitively detect small molecules in the brain because the amount of any given neurochemical is low. Sensors can be designed to change the brightness of an MRI signal in the presence of specific molecules — but it takes a lot of contrast agent to achieve this. If every molecule of contrast agent needs its own target molecule to activate it, low concentrations of the target molecule limit the sensors’ visibility in an MRI scan. “The signal change that you see in the imaging will be very modest,” Jasanoff says. “It won’t let us detect physiological events.”

The Jasanoff team’s new sensors, whose development was led by postdoc Sayani Das and graduate student Jacob Cyert Simon, overcome this problem. To generate a greater signal change in response to target molecules, the researchers designed probes in which a single target molecule impacts not one contrast agent, but many.

To achieve this, Das and Simon packaged an MRI contrast agent inside tiny sacs called liposomal nanoparticles. Each nanoparticle is packed with many molecules of gadolinium, a magnetic material that brightens the MRI signal that arises from hydrogen atoms in water. Inside their protective sacs, gadolinium has no effect on MRI signal, unless water molecules can easily get in and out.

Das and Simon built water channels into the walls of their gadolinium-filled nanoparticles, engineering them so that their opening depends on the presence or absence of a target molecule. When the channels open, more water enters and the gadolinium brightens the local MRI signal, lighting up that spot in a scan.

The researchers call their target-responsive sensors liposomal nanoparticle reporters, or LisNRs (pronounced “listeners”). They designed LisNRs that let water in only in the presence of their target molecule. The water channels in these nanoparticles stay blocked until they encounter their target, which can knock aside a channel-blocking bit of protein.

Once the channel blocker is displaced, water enters and MRI signal brightens. They also made LisNRs that dim the MRI signal in the presence of the molecule they are designed to detect. These have a channel that stays open until the target molecule comes along and blocks it, keeping water out. Jasanoff lab members Vinay Sharma, Samira Abozeid, and Gregory Thiabaud played key roles in understanding and optimizing these interactions, and collaborators in the laboratory of Masayuki Inoue at the University of Tokyo helped the group engineer channels with higher potency.

In experiments led by postdoc Miranda Dawson, Jasanoff’s team used their LisNRs to detect a molecule called biotin in the brains and bodies of living rats, illustrating the probe’s amplifying effects. “We showed that we could detect micromolar-scale levels of biotin with about tenfold greater sensitivity than we would have if we’d used a more conventional, one-to-one type sensing approach,” Jasanoff says. He adds that the team’s modeling suggests that with further development, they may be able to achieve even greater sensitivity gains.

The group showed that the new sensors can be delivered systemically, reaching various organs and spreading throughout the brain. This makes them promising tools for brain-wide imaging, as well as imaging targets in the peripheral nervous system or other tissues.

A next step will be engineering LisNRs that respond to the specific neurochemicals that Jasanoff and his team hope to study. “There are something like 100 neurochemicals in the brain that we’d love to detect, in principle,” he says. They’ll start with dopamine and glutamate — two important and relatively abundant molecules that mediate communications between neurons.

This research, including support for postdoctoral fellows and graduate students involved in the work, was funded, in part, by Lore Harp McGovern, the Yang Tan Collective at MIT, the K. Lisa Yang Brain-Body Center at MIT, the Hock E. Tan and K. Lisa Yang Center for Autism Research at MIT, and the K. Lisa Yang and Hock E. Tan Center for Molecular Therapeutics at MIT.

MIT students study plasma physics beneath Alaska’s aurora

Lauren Bandklayder | Plasma Science and Fusion Center — Fri, 22 May 2026 09:00:00 -0400

For many graduate students, waking up at noon after a 4 a.m. bedtime is a sign of a night well spent. For a group of MIT students, it was simply the start of their workday — timed not to the sun, but to the aurora.

Their goal was simple: to study plasma phenomena using the aurora borealis as a natural laboratory. The process, less so; working largely in darkness in Fairbanks, Alaska, the students conducted experiments in temperatures that dipped as low as -25 degrees Fahrenheit, using red headlamps for visibility. The sun set before 3 p.m., and even at its warmest, temperatures barely reached 20 F.

The aurora provides a rare opportunity to observe plasma behavior directly, as charged particles that interact with Earth’s magnetic field produce visible, large-scale structures in the night sky. As Fairbanks is situated beneath a region of especially frequent auroral activity, it is one of the most reliable places in the world to observe these phenomena, though the conditions come with real constraints.

For one thing, the extreme cold directly impacted the instrumentation. “Our laptops went from full battery to nearly empty in 10 minutes because of the cold,” says Leonardo Corsaro, a PhD student in physics at the Plasma Science and Fusion Center (PSFC) at MIT. “We were trying to transfer data as fast as possible before everything shut down; it was a race against time!”

The challenges extended beyond the cold itself. “The cold can be managed,” says Leon Nichols, a PhD student in physics at PSFC. “With good planning, you can stay comfy in -20 F. The real difficulty was movement when deploying cameras far away from the roads. Walking through thick snow can burn up to 900 calories in an hour. We used cross-country skis to access some of the more remote terrain that would have taken hours to reach otherwise.”

But the conditions were more than worth it: During their time in Alaska, the group witnessed the strongest solar storm in the past two decades, bringing the aurora to life in ways few will ever experience. “It felt like we were the only ones there,” Sydney Menne, a PhD student in nuclear science and engineering, recounts, “removed from the Earth and just entirely surrounded by the aurora, fully immersed in it.”

The team was granted access to observation facilities at Poker Flat Research Range through the University of Alaska Fairbanks Geophysical Institute. Over the course of the trip, students deployed multiple all-sky camera systems across distances of up to 100 miles, enabling simultaneous observations of auroral structures from different locations. These cameras, which capture 360-degree images of the night sky, were paired with magnetometers to correlate visual auroral features with changes in Earth’s magnetic field.

By combining spatially distributed imaging with magnetic field measurements, the team aimed to capture how auroral structures change across space, with the long-term goal of supporting three-dimensional reconstructions of the aurora. This year’s campaign also expanded the measurements beyond imaging, using muon detectors to explore possible correlations between visual auroral activity, magnetic field changes, and particle detections, offering a potential window into how high-energy particles in the upper atmosphere relate to visible auroral activity.

Despite decades of study, many aspects of the aurora remain poorly understood, and each observation offers an opportunity to better characterize the behavior of plasma in near-Earth space. The team also observed a pulsating aurora, a relatively rare phenomenon in which strips of light stretching across the sky blink on and off multiple times per second. By combining instruments not traditionally applied to these problems and deploying low-cost systems at scale, the team is exploring new approaches to studying these phenomena. Insights from these observations can help improve our understanding of space weather, including how solar activity affects satellites, communications systems, and power infrastructure on Earth.

For some participants, the experience reshaped how they think about plasma physics itself. Corsaro explains, “In my research, it is easy to associate these phenomena with colorful plots and simulations, losing touch with the physical process. Seeing structures in the aurora, electric currents and flows forming and shifting overhead, brought a sense of reality to those concepts, and served as a reminder that real plasmas are far less neat and intuitive than theory suggests.”

The experience is part of a broader effort. This group of students represented the third iteration of the Geophysical Plasma Observation Expedition (GPOE), a project involving MIT students from the Plasma Science and Fusion Center, along with collaborating departments, that sends a cohort to Fairbanks, Alaska, each year. Faculty members now provide support for the expedition, while continuity is maintained through its student-driven structure, with each cohort including a mix of returning and new participants. The expedition is organized and led entirely by students and operates on an intensive, compressed timeline. Students are responsible not only for data collection, but also for instrument design, site selection, logistics, and post-processing, completing a full research cycle within a matter of months.

This year’s cohort included graduate students Leonardo Corsaro and Leon Nichols of PSFC; Sydney Menne of the Department of Nuclear Science and Engineering; and Noah Wolfe and Oleksandra “Sasha” Lukina of the Laser Interferometer Gravitational-Wave Observatory (LIGO) Laboratory and the MIT Kavli Institute for Astrophysics and Space Research. The group was accompanied by Professor Matthew Evans, the MathWorks Professor of Physics at MIT, who is affiliated with the LIGO Laboratory and the Kavli Institute.

“This is an opportunity to go from concept to data analysis in just a few months,” says John Ball, a PhD student in nuclear science and engineering at PSFC. “That kind of compressed scientific cycle is rare, especially in our field.”

The program itself has relatively recent and somewhat unusual origins. It began in 2023, when graduate student Shon Mackie, frustrated by the lack of hands-on plasma diagnostic opportunities, noticed the solar cycle was approaching its peak and saw an opportunity to study plasma phenomena more directly. He drafted a short proposal to PSFC leadership, and the response from then-Director Dennis Whyte was two lines: “Sounds cool, literally! PSFC will fund this.”

Since its launch in 2023, GPOE has evolved from a single-camera effort into a multi-instrument, multi-site campaign with growing participation, with each cohort building on the work of previous years by refining instrumentation, expanding observational coverage, and improving data collection strategies.

This hands-on, student-driven approach has also created opportunities to extend the experience beyond MIT. In 2024, the program expanded to include a new outreach collaboration with the MIT Museum and the MIT Nord Anglia Collaboration, bringing approximately 65 high school students from around 20 schools worldwide to MIT to help design and build components of the all-sky camera systems used in the field. Working within a set of technical constraints, students developed and tested designs, ultimately producing 13 cameras that were deployed during the Alaska expedition.

The program has also begun to produce results beyond the expedition itself. Students have presented their work at major conferences, including the American Geophysical Union, and published findings in peer-reviewed journals such as Earth and Space Science. The group’s low-cost all-sky camera and magnetometer design is now being adopted by other research teams and community science initiatives, extending its impact beyond MIT.

Beyond its scientific goals, participants emphasized the broader impact of the experience.

“Standing outside at midnight in Alaska, staring up at sheets of glowing plasma stretching thousands of kilometers across the sky, really brings home just how small and delicate our own place in the universe is,” says Ball.

As the program continues to grow, students hope to expand both its technical capabilities and its reach, including more permanent instrumentation and expanding outreach partnerships. For many involved, the expedition represents not just a research opportunity, but a reminder of the scale and immediacy of the phenomena they study.

“Science is an adventure,” Corsaro says. “This kind of work reminds you why you became a scientist in the first place.”

The rules neurons follow to make sense of what we see

David Orenstein | The Picower Institute for Learning and Memory — Thu, 21 May 2026 16:50:00 -0400

Even in the primary visual cortex, a brain region named for its specialized role in processing basic features of what the eyes see, not every neuron ends up answering the call to process properties of visual input. Maybe that’s because each neuron receives a wide variety of inputs via thousands of circuit connections, or “synapses,” and has to opt to respond to the visual information versus something else. In a new study in mice, neuroscientists at The Picower Institute for Learning and Memory at MIT reveal how neurons that perform visual processing bring order to this input to get the job done.

Neuroscientists are keenly interested in what inputs, from among so many choices, will compel neurons to participate in the brain’s computations and functions, says senior author Mriganka Sur, Newton Professor of Neuroscience in the Picower Institute and MIT’s Department of Brain and Cognitive Sciences. Neurons ultimately participate in brain circuits by “firing” an electrical action potential.

“The configuration of inputs, the kind of organization, the assembly of neurons that modulate each other to generate an action potential is the essence of how brain circuits process information,” Sur says. “These (visual cortex) cells are a microcosm of this very profound and big picture of neuroscience.”

In the open-access study in iScience, led by postdoc Kyle Jenks, the research team achieved their findings by meticulously imaging how not only neurons’ cell bodies, but also their individual synapses, formed on protrusions known as dendritic spines, responded as mice viewed moving images. They did this imaging for not only visually responsive neurons, but also for unresponsive neurons that nevertheless have visually responsive spines. That allowed them to analyze many key properties that might influence where a particular synapse forms, and how it influences responses at the cell body.

“This pulls together a lot of things that have been looked at in isolation and looks at them in one collective paper,” Jenks says. “We can compare how the neuron and the spines on that neuron respond to the same stimuli, and we can do this for both visually responsive and unresponsive neurons.”

In visual cortex layer 2/3, Jenks and the team genetically engineered neurons such that their individual dendritic spines would glow when surges of calcium indicated increased activity by the synapses on the spines. The scientists did the same for the cell body, or “soma,” to keep track of how the cell responded and even signaled its overall responses back out to the synapses. This way, as the mice watched black and white gratings at varying angles drift by their eyes in different directions, the scientists could keep track of each spine’s and each cell’s overall response to that patterned visual input.

In all, they tracked 11 neurons that responded to the visual input and 11 others that seemingly ignored it. That enabled them to find several rules:

Distance from the soma matters: On cells that responded to visual input, the responses of individual spines were much more likely to correlate with the activity of the soma the closer the spine was to the soma. In the same vein, the soma’s signal back out to spines, which is believed to influence the spines’ alignment with the soma’s preferences, was more likely to be detectable closer to the soma than farther away.

Local clustering: On neurons that responded to visual input, spines formed distinct little enclaves of correlated responses with each other. Specifically, spines within 5 microns (five one-millionths of a meter) acted in concert. But then, right outside that 5-micron boundary, spines were less likely than chance to join in that activity. Sur speculates that these isolated pockets of activity sharpened the response from each enclave.

“Apical” vs. “basal:” The neurons the team studied have two distinct kinds of dendrites. Apical dendrites, which are very long and protrude from the top, or “apex,” of the neuron, tend to get a wide variety of inputs from across the cortex. Basal dendrites, which are shorter and extend out from the bottom, typically get more raw visual input. While basal dendrites indeed received more visual input than apical dendrites overall, Jenks found that apical dendrites on visually responsive neurons had significantly more visually responsive spines than those on non-responsive neurons. And both types of dendrites equally obeyed the rules above about distance from the soma.

Orientation selectivity matters most: Jenks, Sur, and the team used statistical modeling to determine which of many factors (the stimulus selectivity, reliability of the response, a spine’s distance from the soma, apical versus basal, etc.) most explained how correlated a spine’s responsiveness was with that of the soma. By a wide margin, how selective a spine was to the orientation of its preferred grating was the most important single factor.

“Our results reveal that synaptic inputs to excitatory layer 2/3 neurons in mouse (visual cortex) are not randomly arranged, but organized and distributed in a manner that correlates with multiple factors including somatic responsiveness, somatic tuning, branch type, distance from the soma, local correlations, and stimulus selectivity,” the researchers wrote.

The team’s findings can help advance studies of vision in the brain in multiple ways, Jenks and Sur say. Certain genetic mutations that affect how neurons connect in circuits can affect visual cortex neurons and vision, Sur says. Documenting these rules provides researchers with a baseline to compare against when examining the effects of such mutations. Jenks adds that the findings could inform efforts to model how neurons integrate synaptic inputs in their computations.

In addition to Sur and Jenks, the paper’s other authors are Gregg Heller, Katya Tsimring, Kendyll Martin, Asrah Rizvi, and Jacque Pak Kan Ip.

The National Institutes of Health, the Simons Foundation Autism Research Initiative, and the Freedom Together Foundation provided support for the study.

Some democracies are struggling to ensure safe drinking water

Peter Dizikes | MIT News — Thu, 21 May 2026 13:30:00 -0400

About 2 billion people — just under a quarter of the world’s population — lack regular access to clean drinking water. And roughly 800,000 people annually die from illnesses associated with unsanitary water.

Drinking water access is a fundamental problem for human and economic development. The U.N., for instance, highlighted the issue in its Sustainable Development Goals of 2015, an ambitious 17-point agenda that specified safe drinking water as a basic global aim.

Past research shows that democracies, in comparison to other forms of government, tend to be more successful at delivering this kind of public good, which benefits a large portion of the population. This is likely due to accountability measures that include elections, greater transparency, and more freedom in civil society.

But now a study led by an MIT professor shows that across nearly 100 countries with developing economies, that dynamic has become more complex in the 21st century. While democracies are slightly ahead of non-democracies when it comes to providing at least some water, they have been falling behind when it comes to ensuring that there is safe water on tap.

“Among low- and middle-income countries, which have not done as well economically, we found there wasn’t really a big difference between democracies and non-democracies in the provision of what is called basic drinking water,” says MIT political scientist Evan Lieberman, co-author of a new paper detailing the results. “But for safe drinking water, we found, surprisingly, that democratic countries were becoming less good at extending access.”

While the study does not pinpoint the precise reasons for this, it suggests a lens for viewing the problem. Democracies tend to be better at delivering visible public goods, the kinds of things citizens can literally see — like infrastructure that delivers water. But the difference between safe and unsafe water is not necessarily visible and obvious, so public officials may not be as responsive.

“This is likely a big part of the equation, that the invisibility of safe water makes it a less compelling public good for politicians,” says Lieberman, the Total Professor of Political Science and Contemporary Africa, and director of MIT’s Center for International Studies.

The paper, “Beyond the tap: The limited value of democracy for delivering universal safe water access in low- and middle-income countries,” is published in the journal World Development. The authors are Lieberman, and Naomi Tilles, a doctoral student in political science at Stanford University.

Seeing is believing

To conduct the study, the scholars analyzed drinking water data recorded by the World Health Organization/UNICEF Joint Monitoring Programme. That provides information for basic availability to water, defined as access to an improved water source with no more than 30 minutes of collection time; and access to safe drinking water, defined as an improved water source that is available on premises, available when needed, and free from potentially disease-producing contaminants, which range from fecal matter to harmful chemicals.

Examining 96 low- and middle-income countries, the researchers looked at a variety of measures pertaining to its democratic or non-democratic features, and ran 39,000 regressions to see how the form of government related to its provision of water. Overall, Lieberman and Tilles found that democratic governance is modestly associated with an increase in the basic availability of water, compared to non-democracies. However, the effect is not particularly robust.

The good news is that between 2000 and 2024, 81 of the 90 countries with data available in both years made gains in safe drinking water access. However, democratic countries have been less successful than their non-democratic counterparts in advancing the goal of achieving universal access.

“Moreover, the gap between democracies and non-democracies seems to be getting a little bit larger over time,” Lieberman observes.

Because the study is focused on establishing the overall empirical situation, the scholars do not claim to have determined why this trend has been emerging. Many newer democracies have struggled to establish high-functioning governance in some regions, which may influence their overall results.

More broadly, Lieberman suggests, visibility matters. Past scholarship has shown that democracies perform relatively well in delivering visible public goods, especially in countries with little information in the public sphere. Delivering water generates attention for politicians in a way that keeping water safe does not.

“Politicians may figure out they should do things citizens like, to stay in office, such as bringing water to an area,” Lieberman says. “You can have a ribbon-cutting ceremony, and people feel it really happened. But water quality is often invisible.

It’s a more difficult challenge to ensure safe water: You have to do testing, prevent people from polluting, and you may need to treat the water.”

In any case, Lieberman notes, “Given what we find, what is clear is that the incentives are not aligned under the current systems for advancing safe-water access within all democracies. That provides opportunities for human agency to create incentives for citizens, nongovernment agencies, and governments to do what is needed.”

Development for all

Lieberman comes to the topic of water access as an expert on African politics. His most recent book, “Until We Have Won Our Liberty” (Princeton University Press, 2022), examines the vicissitudes of South African democracy. In the book and in general, he suggests that democracy is the most viable path toward development with “dignity,” meaning economic growth accompanied by liberties and equal treatment under the law.

“I think democracy provides dignified development, by granting people recognition and participation, and that’s an extremely valuable thing,” Lieberman says.

Still, when it comes to the performance of many countries with regard to safe water, he says, “I think we just need to be clear-eyed about real problems.”

In some countries, he suggests, the time horizon of elected officials may also be relatively short-term, and they may be more oriented toward simpler problems than water safety. At the same time, other members of society need to find ways to make water safety a bigger issue in the eyes of the public.

“There are important lessons for democracies to learn, and citizens in civil society who are aware of this challenge need to figure out ways to get people to care about it, to recognize the connection between illness and unsafe water, and to use political campaigns to advance their longer-term interests,” Lieberman says.

Overall, he adds, “There is something intrinsically important about democratic government. Then the question becomes how to make it work better to deliver really important outcomes like safe water.”

Technology usually creates jobs for young, skilled workers. Will AI do the same?

Peter Dizikes | MIT News — Thu, 21 May 2026 00:00:00 -0400

At any given time, technology does two things to employment: It replaces traditional jobs, and it creates new lines of work. Machines replace farmers, but enable, say, aeronautical engineers to exist. So, if tech creates new jobs, who gets them? How well do they pay? How long do new jobs remain new, before they become just another common task any worker can do?

A new study of U.S. employment led by MIT labor economist David Autor sheds light on all these matters. In the postwar U.S., as Autor and his colleagues show in granular detail, new forms of work have tended to benefit college graduates under 30 more than anyone else.

“We had never before seen exactly who is doing new work,” Autor says. “It’s done more by young and educated people, in urban settings.”

The study also contains a powerful large-scale insight: A lot of innovation-based new work is driven by demand. Government-backed expansion of research and manufacturing in the 1940s, in response to World War II, accounted for a huge amount of new work, and new forms of expertise.

“This says that wherever we make new investments, we end up getting new specializations,” Autor says. “If you create a large-scale activity, there’s always going to be an opportunity for new specialized knowledge that’s relevant for it. We thought that was exciting to see.”

The paper, “What Makes New Work Different from More Work?” is forthcoming in the Annual Review of Economics. The authors are Autor; Caroline Chin, a doctoral student in MIT’s Department of Economics; Anna M. Salomons, a professor at Tilburg University’s Department of Economics and Utrecht University’s School of Economics; and Bryan Seegmiller PhD ’22, an assistant professor at Northwestern University’s Kellogg School of Management.

And yes, learning about new work, and the kinds of workers who obtain it, might be relevant to the spread of artificial intelligence — although, in Autor’s estimation, it is too soon to tell just how AI will affect the workplace.

“People are really worried that AI-based automation is going to erode specific tasks more rapidly,” Autor observes. “Eroding tasks is not the same thing as eroding jobs, since many jobs involve a lot of tasks. But we’re all saying: Where is the new work going to come from? It’s so important, and we know little about it. We don’t know what it will be, what it will look like, and who will be able to do it.”

“If everyone is an expert, then no one is an expert”

The four co-authors also collaborated on a previous major study of new work, published in 2024, which found that about six out of 10 jobs in the U.S. from 1940 to 2018 were in new specialties that had only developed broadly since 1940. The new study extends that line of research by looking more precisely at who fills the new lines of work.

To do that, the researchers used U.S. Census Bureau data from 1940 through 1950, as well as the Census Bureau’s American Community Survey (ACS) data from 2011 to 2023. In the first case, because Census Bureau records become wholly public after about 70 years, the scholars could examine individual-level data about occupations, salaries, and more, and could track the same workers as they changed jobs between the 1940 and 1950 Census enumerations.

Through a collaborative research arrangement with the U.S. Census Bureau, the authors also gained secure access to person-level ACS records. These data allowed them to analyze the earnings, education, and other demographic characteristics of workers in new occupational specialties — and to compare them with workers in longstanding ones.

New work, Autor observes, is always tied to new forms of expertise. At first, this expertise is scarce; over time, it may become more common. In any case, expertise is often linked to new forms of technology.

“It requires mastering some capability,” Autor says. “What makes labor valuable is not simply the ability to do stuff, but specialized knowledge. And that often differentiates high-paid work from low-paid work.” Moreover, he adds, “It has to be scarce. If everyone is an expert, then no one is an expert.”

By examining the census data, the scholars found that back in 1950, about 7 percent of employees had jobs in types of work that had emerged since 1930. More recently, about 18 percent of workers in the 2011-2023 period were in lines of work introduced since 1970. (That happens to be roughly the same portion of new jobs per decade, although Autor does not think this is a hard-and-fast trend.)

In these time periods, new work has emerged more often in urban areas, with people under 30 benefitting more than any other age category. Getting a job in a line of new work seems to have a lasting effect: People employed in new work in 1940 were 2.5 times as likely to be in new work in 1950, compared to the general population. College graduates were 2.9 percentage points more likely than high school graduates to be engaged in new work.

New work also has a wage premium, that is, better salaries on aggregate than in already-existing forms of work. Yet as the study shows, that wage premium also fades over time, as the particular expertise in many forms of new work becomes much more widely grasped.

“The scarcity value erodes,” Autor says. “It becomes common knowledge. It itself gets automated. New work gets old.”

After all, Autor points out, driving a car was once a scarce form of expertise. For that matter, so was being able to use word-processing programs such as WordPerfect or Microsoft Word, well into the 1990s. After a while, though, being able to handle word-processing tools became the most elementary part of using a computer.

Back to AI for a minute

Studying who gets new jobs led the scholars to striking conclusions about how new work is created. Examining county-level data from the World War II era, when the federal government was backing new manufacturing in public-private partnerships throughout the U.S., the study shows that counties with new factories had more new work, and that 85 to 90 percent of new work from 1940 to 1950 was technology-driven.

In this sense there was a great deal of demand-driven innovation at the time. Today, public discourse about innovation often focuses on the supply side, namely, the innovators and entrepreneurs trying to create new products. But the study shows that the demand side can significantly influence innovative activity.

“Technology is not like, ‘Eureka!’ where it just happens,” Autor says. “Innovation is a purposive activity. And innovation is cumulative. If you get far enough, it will have its own momentum. But if you don’t, it’ll never get there.”

Which brings us back to AI, the topic so many people are focused on in 2026. Will AI create good new jobs, or will it take work away? Well, it likely depends how we implement it, Autor thinks. Consider the massive health care sector, where there could be a lot of types of tech-driven new work, if people are interested in creating jobs.

“There are different ways we could use AI in health care,” Autor says. “One is just to automate people’s jobs away. The other is to allow people with different levels of expertise to do different tasks. I would say the latter is more socially beneficial. But it’s not clear that is where the market will go.”

On the other hand, maybe with government-driven demand in various forms, AI could get applied in ways that end up boosting health care-sector productivity, creating new jobs as a result.

“More than half the dollars in health care in the U.S. are public dollars,” Autor observes. “We have a lot of leverage there, we can push things in that direction. There are different ways to use this.”

This research was supported, in part, by the Hewlett Foundation, the Google Technology and Society Visiting Fellows Program, the NOMIS Foundation, the Schmidt Sciences AI2050 Fellowship, the Smith Richardson Foundation, the James M. and Cathleen D. Stone Foundation, and Instituut Gak.

Four from MIT named 2026 Searle Scholars

Wed, 20 May 2026 16:00:00 -0400

MIT scientists Sven Dorkenwald and Whitney Henry have been named 2026 Searle Scholars, an award given annually to 15 exceptional early-career researchers in the fields of biomedical sciences and chemistry. Dorkenwald is an assistant professor of brain and cognitive sciences and an investigator at the McGovern Institute for Brain Research. Henry is the Robert A. Swanson (1969) Career Development Professor of Life Sciences and an intramural faculty member at the Koch Institute for Integrative Cancer Research.

In addition, MIT alumni Irene Kaplow ’10 and Jared Mayers PhD ’15 were also honored.

Chosen by a scientific advisory board, Searle Scholars are considered among the most creative young researchers pursuing high-risk/high-reward research. The Searle Scholars Program is funded through the Searle Funds at The Chicago Community Trust and administered by Kinship Foundation. Each scholar will each receive $450,000 in flexible funding to support their work over the next three years.

Sven Dorkenwald

Sven Dorkenwald is a computational neuroscientist investigating the organizational principles of neuronal circuits. The synaptic connectivity of neurons, their connectome, is fundamental to how networks of neurons function. Dorkenwald develops computational and collaborative tools to map, analyze, and interpret synapse-resolution connectomes. His work has led to large connectomic reconstructions of the fruit fly brain and parts of mammalian brains. He uses these connectomes to investigate the architecture of neuronal circuits and how their structure supports complex computations.

“As I establish my new lab, the Searle Scholars Award will help us launch ambitious projects and set our long-term scientific direction,” says Dorkenwald. “I am deeply grateful for the support from the Kinship Foundation and look forward to interacting with this amazing cohort of Searle Scholars.”

Dorkenwald joined the faculty of MIT in 2026 as an assistant professor in the Department of Brain and Cognitive Sciences and an investigator at the McGovern Institute. He earned a BS in physics and an MS in computer engineering from the University of Heidelberg, followed by a PhD in computer science and neuroscience at Princeton University in 2023 under the mentorship of Sebastian Seung and Mala Murthy. Dorkenwald completed his postdoctoral training as a Shanahan Research Fellow at the Allen Institute and the University of Washington, while serving as a visiting faculty researcher at Google Research.

Whitney Henry

Whitney Henry investigates the potential of ferroptosis, an iron-dependent form of cell death, for developing novel therapies that target subpopulations of cancer cells that are highly metastatic, therapy-resistant, and therefore critical instigators of tumor relapse. Her research is focused on uncovering the molecular factors influencing ferroptosis susceptibility, investigating its effects on the tumor microenvironment, and developing innovative methods to manipulate ferroptosis resistance in living organisms, drawing from functional genomics, metabolomics, bioengineering, and a range of in vitro and in vivo models.

“I am incredibly grateful to the Kinship Foundation for supporting our research and giving us the freedom to ask bold, curiosity-driven scientific questions,” says Henry. “This support allows us to pursue ambitious ideas, take creative risks, and embark on new research directions.”

Henry joined the MIT faculty in 2024 as an assistant professor in the Department of Biology and a member of the Koch Institute, and is currently an HHMI Freeman Hrabowski Scholar. She received her bachelor's degree in biology with a minor in chemistry from Grambling State University and her PhD from Harvard University. Following her doctoral studies, she worked in the lab of Robert Weinberg at the Whitehead Institute for Biomedical Research and was supported by fellowships from the Jane Coffin Childs Memorial Fund for Medical Research and the Ludwig Center at MIT.

Alumni also honored

Irene Kaplow ’10, a graduate of the MIT Department of Mathematics, is an assistant professor in the Department of Biology and the Ray and Stephanie Lane Computational Biology Department at Carnegie Mellon University. Her selection as a Searle Scholar is for “deciphering transcriptional regulatory mechanisms underlying mammalian dietary phenotype evolution and their relationships to transcriptional regulatory responses to changes in diet.”

Jared Mayers PhD ’15, who earned his doctorate from the MIT Department of Biology, is an assistant professor at the Fred Hutchinson Cancer Center at the University of Washington. His selection as a Searle Scholar is for “a reverse-translational framework to decipher metabolic vulnerabilities of bacterial pathogens.”

The Haystack 37m Telescope: A new era of astrophysical research

Nancy Wolfe Kotary | MIT Haystack Observatory — Tue, 19 May 2026 13:25:00 -0400

The Haystack 37m Telescope has been a landmark in radio astronomy and radar studies of the solar system since its first light in 1964. Over the following four decades, it supported NASA's Apollo landings on the moon, made planetary radar maps of the surface of Venus, contributed to experimental tests of Einstein's general relativity, supported the development of VLBI, and conducted foundational studies of quasars and star-forming regions.

Recently, the Haystack 37m Telescope — a 37-meter radio and millimeter-wavelength antenna at MIT Haystack Observatory in Westford, Massachusetts — made its return to front-line astronomical research following an extended period of system upgrades. These observations reconnect this instrument with its long tradition of scientific discovery and open a new chapter.

On Dec. 8, 2025, Haystack scientists observed the supermassive black hole system at the center of the galaxy Messier 87 (M87) using a technique called very long baseline interferometry (VLBI) that links telescopes across continents to achieve extraordinary resolution. These observations mark the return of one of America's most storied radio telescopes to its historical scientific and educational mission.

The observations targeted the powerful jet of energy and matter launched from M87’s central black hole, M87*. This jet, driven by a black hole six-and-a-half billion times the mass of our sun, extends thousands of light years into intergalactic space and is one of the most energetic phenomena in the known universe.

Previous international campaigns, namely those led by the Event Horizon Telescope, have imaged the black hole's immediate “shadow.” The Haystack 37m Telescope observations, performed in collaboration with the telescopes of the Very Long Baseline Array (VLBA) and the Greenland Telescope (GLT), help to probe the larger-scale structure of the jet, investigating how energy is transported far beyond the black hole's vicinity. Understanding this process is central to explaining how supermassive black holes shape the galaxies that surround them.

“The Haystack 37m Telescope’s exceptional sensitivity enables the intercontinental telescope array to detect faint emission from around the distant M87* black hole,” says Paul Tiede, principal investigator of the M87 study. “In tandem with the GLT and the VLBA, Haystack is helping create the first multifrequency movies of M87*’s faint jet, greatly improving our understanding of black hole physics.”

The upgraded Haystack 37m Telescope opens multiple new lines of research. At MIT, Saverio Cambioni and Richard Teague of the Department of Earth, Atmospheric and Planetary Sciences (EAPS) plan to use the instrument within MIT’s Planetary Defense Project to measure asteroid sizes and shapes, characterizing objects that could pose a hazard to Earth and deepening our understanding of the solar system’s formation. Associate Professor Brett McGuire of the Department of Chemistry plans to search for complex organic molecules in space, work that speaks to the question of how the chemical precursors to life arise.

“We are thrilled to provide the research community with a powerful telescope at a time where few such instruments are available,” says Jens Kauffmann, principal investigator of the Haystack 37m Telescope Astronomy Program, who uses the telescope to study the formation of stars and their planets. “Even more exciting are the prospects this generates for the next generation of astronomers. Hands-on training opportunities on world-class research telescopes have become exceptionally rare worldwide, and now we can offer this singular advanced workforce development program right here in Massachusetts.”

Student involvement with the Haystack 37m Telescope has already resumed: Undergraduate interns at Haystack Observatory played an active role in developing the telescope’s control systems and data analysis algorithms. This work exemplifies Haystack’s role as a hands-on research and training environment where students contribute directly and gain practical experience with a frontline research instrument.

The return to research-focused observations is the result of more than 10 years of careful, sustained work. From 2010 to 2014, the Haystack 37m Telescope underwent a major upgrade and refurbishment that enhanced its ability to observe at millimeter wavelengths. This work was primarily done to improve the antenna’s capability as a space radar. The dish now primarily serves U.S. government agencies in that capability, and astronomy was temporarily a secondary activity.

But work to restore the telescope's science capability never stopped. Initial support from the National Science Foundation (NSF) in 2015 modernized systems for data analysis and radio signal processing. The first successful engineering-oriented VLBI experiments with the new dish were conducted at the same time. Additional NSF funding in 2019, provided in the context of the Next Generation Event Horizon Telescope (ngEHT) program, enabled a more general and sustained effort to upgrade receiver equipment and computing systems. Support from private donors to Haystack also aided in this longer-term effort.

Several recent developments, particularly in 2025, proved significant. With support from MIT's Jarve Seed Fund for Science Innovation, scientists and engineers removed lingering technical limitations with astronomy systems and expanded the telescope's scientific reach. Other funding for projects led by the Smithsonian Astrophysical Observatory enabled the M87 campaign and commissioning of the next-generation digital back end, a highly advanced signal-processing system developed for the ngEHT. Together, these advances made the December 2025 observations possible. MIT Haystack Observatory is now pursuing support from both private and federal sources for further improvements under the Haystack 37m Telescope Astronomy Program.

“The upgraded Haystack 37m Telescope empowers MIT students and researchers to pursue fundamental questions relating to our origins and our solar system,” says Richard Teague, professor at MIT EAPS. “With privileged access to such a powerful facility, we can undertake ambitious observational programs previously impossible to schedule. This is the beginning of what we expect will be an exciting era of new discoveries with the Haystack 37m Telescope.”

Single-molecule tracker illuminates workings of cancer-related proteins

Leah Eisenstadt | Broad Institute — Tue, 19 May 2026 11:35:00 -0400

Using a powerful single-molecule imaging method they developed, a research team from the Broad Institute of MIT and Harvard has unveiled a dynamic view of how some cancer-related proteins interact in living cells.

The technique relies on highly stable nanoparticle probes that brightly illuminate individual molecules for long periods of time. The researchers used their method to observe, for the first time, individual receptors as they move around the cell membrane, attaching to and then letting go of other receptors to alter signaling within the cell.

Described in the journal Cell, the work demonstrates the method’s potential for investigating other receptors and molecules, and for improved drug screening to better understand the effects of therapeutics on living cells.

“With our photostable probes, we can map out the entire lifespan of these molecules in their native environment and see things that have never been observable before,” says study leader Sam Peng, a Broad Institute core institute member and assistant professor of chemistry at MIT.

Molecular movies

Peng’s method solves a problem with existing contrast agents used in single-molecule tracking, such as dyes. Under the laser light that’s used to excite these dyes, they burn out after a few seconds in a phenomenon known as photobleaching, which means that scientists could only use them to take a few snapshots of cell receptors, and not follow them over the entirety of the signaling process.

For a longer and richer view, Peng’s lab developed long-lasting probes, known as upconverting nanoparticles, which emit signals that remain stable under laser excitation. The nanoparticles contain rare-earth ions that continue to luminescence for minutes, hours, and potentially years. In addition, by altering the type and doses of the ions, scientists can engineer probes emitting in many different colors, enabling tracking of many targets in a single experiment.

In the current study, the researchers aimed to uncover new biology by focusing on the EGFR family of cell receptors, which have been linked to several kinds of cancer. They collaborated with EGFR experts Matthew Meyerson and Heidi Greulich of the Broad’s Cancer Program. They knew that EGFR receptors need to pair up, or “dimerize,” in order to initiate signaling within the cell, but they wanted to learn more about the dynamics of these pairings — what the receptors partner with, how long they stay together, and how they find new partners.

For a better and more sustained look at the receptors, the research team customized their upconverting nanoparticles to tag EGFR and related receptors HER2 and HER3, which are linked to cancer, and used them to track the molecules in living human cells.

A new view of protein pairings

In this study, Peng and his team observed that, when activated with a stimulating molecule, EGFR receptors can pair up and stay dimerized for several minutes, something not observable using traditional dyes. Excessive and prolonged dimerization can lead to too much cell growth and cancer.

A microscopy video shows upconverting nanoparticles tagged to EGFR receptors (labeled pink and green), which track individual receptors as they dimerize. Image courtesy of the researchers.

When the EGFR molecules carried cancer-related mutations, the dimers became more stable, with the more stabilizing mutations linked to more potent cancers in people. In addition, the mutated receptors could form stable dimers even without an external stimulus prompting them to dimerize. The finding helps explain how EGFR mutations can lead to uncontrolled cell growth and cancer, and could inform efforts to target this process therapeutically.

The team discovered several other new and surprising details about how HER2 and HER3 form stable pairings with themselves, which helps illuminate the role of these molecules in related cancers.

When the research team tagged all three receptor types in one experiment, they observed a vibrant scene with receptors navigating the cell surface, finding partners, unpairing, and then finding new partners, over and over again.

Beyond shedding light on EGFR biology, the scientists hope that collaborators in other fields will apply their method to ask new scientific questions about other proteins of interest. “We think this technique could be transformative for studying molecular biology, because it enables dynamic biological processes to be observed with high spatiotemporal resolution over unprecedented timescales,” says Peng.

They are also planning to explore the method’s use in studying the mechanism of drug action, to reveal how potential therapeutics alter individual molecules over time. In addition, they will continue to improve their methods, such as making the probes smaller, brighter, and able to emit more colors.

New research enables a robot to chart a better course

Adam Zewe | MIT News — Tue, 19 May 2026 00:00:00 -0400

In the aftermath of a devastating earthquake, unpiloted aerial vehicles (UAVs) could fly through a collapsed building to map the scene, giving rescuers information they need to quickly reach survivors.

But this remains an extremely challenging problem for an autonomous robot, which would need to swiftly adjust its trajectory to avoid sudden obstacles while staying on course.

Researchers from MIT and the University of Pennsylvania developed a new trajectory-planning system that tackles both challenges at once. Their technique enables a UAV to react to obstacles in milliseconds while staying on a smooth flight path that minimizes travel time.

Their system uses a new mathematical formulation that ensures the robot travels safely to its destination along a feasible path, and that is less computationally intensive than other techniques. In this way, it generates smoother trajectories faster than state-of-the-art methods.

The trajectory planner is also efficient enough for real-time flight using only the robot’s onboard computer and sensors.

Named MIGHTY, the open-source system does not require proprietary software packages that can cost hundreds of thousands of dollars. It could be more readily deployed in a wider variety of real-world settings.

In addition to search-and-rescue, MIGHTY could be utilized in applications like last-mile delivery in urban spaces, where UAVs need to avoid buildings, wires, and people, or in industrial inspection of complex structures, such as wind turbines.

“MIGHTY achieves comparable or better performance using only open-source tools, which means any researcher, student, or company — anywhere in the world — can use it freely. By removing this cost barrier, MIGHTY helps democratize high-performance trajectory planning and opens the door for a much broader community to build on this work,” says Kota Kondo, an aeronautics and astronautics graduate student and lead author of a paper on this trajectory planner.

Kondo is joined on the paper by Yuwei Wu, a graduate student at the University of Pennsylvania; Vijay Kumar, a professor at UPenn; and senior author Jonathan P. How, a Ford professor of aeronautics and astronautics and a principal investigator in the Laboratory for Information and Decision Systems (LIDS) and the Aerospace Controls Laboratory (ACL) at MIT. The research appears in IEEE Robotics and Automation Letters.

Overcoming trade-offs

When Kondo was a child, the Fukushima Daiichi nuclear accident occurred following the Great East Japan Earthquake. With school cancelled, Kondo was stuck at home and watched the news every day as workers explored and secured the reactor site. Some workers still had to enter hazardous areas to contain the damage and assess the situation, exposing them to high doses of radioactive material.

“I became passionate about creating autonomous robots that can go into these dynamic and dangerous situations, then come back and report to humans who stay out of harm’s way,” Kondo says.

This task requires a strong trajectory planner, which is software that decides the path a robot should follow to safely get from point A to point B.

But many existing systems force tradeoffs that limit performance.

While some commercial systems can rapidly generate smooth trajectories, they can cost hundreds of thousands of dollars. Open-source alternatives often underperform compared to commercial solvers or are difficult to use.

With MIGHTY, Kondo and his colleagues developed an open-source system that produces high-quality, smooth trajectories while reacting to obstacles in real-time, and which runs fast enough for flight using only onboard components.

To do this, they overcame a key challenge that limits many open-source systems.

These methods usually estimate how long it will take the robot to get from point A to point B as a first step. From that fixed estimation of travel time, the planner finds the best path to reach the destination.

While using a fixed travel time allows the planner to rapidly generate a trajectory, it has drawbacks. For one, if the UAV must go far out of its way to avoid obstacles, it could be forced to crank up the speed to meet the fixed travel-time budget. This makes it harder to avoid sudden hazards.

A MIGHTY method

Instead, MIGHTY uses a mathematical technique, called a Hermite spline, that optimizes the travel time and flight path together, in a single step, to form a smooth trajectory that can be precisely controlled.

“Optimizing the spatial and temporal components together gets us better results, but now the optimization becomes so much bigger that it is harder to solve in a feasible amount of time,” Kondo says.

The researchers used a clever technique to reduce this computational overhead.

Instead of generating a trajectory from scratch each time, MIGHTY makes an initial guess of a trajectory. Then it refines the trajectory through an iterative optimization, using a map of the scene generated by the UAV’s lidar sensors.

“We can make a decent guess of what the trajectory should be, which is a lot faster than generating the entire thing from nothing,” Kondo says.

This enables MIGHTY to react in real-time to unknown obstacles while keeping the trajectory smooth and minimizing travel time. The system utilizes the UAV’s onboard components, which is important for applications where a robot might travel far from a base station.

In simulated experiments, MIGHTY needed only about 90 percent of the computation time required by state-of-the-art methods, while safely reaching its destination about 15 percent faster than these approaches.

When they tested the system on real robots, it reached a speed of 6.7 meters per second while avoiding every obstacle that appeared in its path.

“With MIGHTY, everything is integrated in one piece. It doesn’t need to talk to any other piece of software to get a solution. This helps us be even faster than some of the commercial solvers,” Kondo says.

In the future, the researchers want to enhance MIGHTY so it can be used to control multiple robots at once and conduct more flight experiments in challenging environments. They hope to continue improving the open-source system based on user feedback.

“MIGHTY makes an important contribution to agile robot navigation by revisiting the trajectory representation itself. Hermite splines have already been successfully used in visual simultaneous localization and mapping, and it is nice to see their advantages now being exploited for trajectory planning in mobile robots. By enabling joint optimization of path geometry, timing, velocity, and acceleration while retaining local control of the trajectory, MIGHTY gives robots more freedom to compute fast, dynamically feasible motions in cluttered environments,” says Davide Scaramuzza, professor and director of the Robotics and Perception Group at the University of Zurich, who was not involved with this research.

This research was funded, in part, by the United States Army Research Laboratory and the Defense Science and Technology Agency in Singapore.

Language development in the brain

Jennifer Michalowski | McGovern Institute for Brain Research — Mon, 18 May 2026 15:15:00 -0400

The brain’s capacity to use and understand language expands rapidly in the first years of life, as babies start to make sense of the words they hear and eventually begin to piece together sentences of their own. The language-processing parts of the brain that make this possible continue to evolve in older children, as they expand their vocabularies and learn to use language more flexibly.

MIT brain researchers have captured snapshots of the developing language-processing network in brain scans of hundreds of children and adolescents. Their data, reported May 16 in the journal Nature Communications, show that the network continues to mature, becoming better integrated and increasingly responsive until around age 16. But they also found that a key feature of the adult language network is established early on: its localization in the left side of the brain.

Language lateralization

It is well known that using language is mostly the job of the left hemisphere. As adults, we call on the language-processing regions there when we read, write, speak, or listen to others talk. But there was some question as to whether this left lateralization is established early in life, or might instead emerge as the language network matures, with both sides of the brain contributing to language in childhood.

To find out, researchers needed to see young brains in action — and several MIT labs had collected exactly the right kind of data. Groups led by Evelina Fedorenko, an associate professor of brain and cognitive sciences; John Gabrieli, the Grover Hermann Professor of Health Sciences and Technology; and Rebecca Saxe, the John W. Jarve (1978) Professor of Brain and Cognitive Sciences, teamed up to share brain scans from children, adolescents, and adults and compare how their brains responded to language. Fedorenko, Gabrieli, and Saxe are also investigators at the McGovern Institute for Brain Research.

In studies aimed at better understanding a variety of cognitive functions and developmental disorders, the three teams had all collected functional MRI data while subjects participated in “language localizer” tasks — an approach the Fedorenko lab developed to map the language-processing network in a person’s brain. By monitoring brain activity with functional MRI as people engage in both language tasks and non-linguistic tasks, researchers can identify parts of the brain that are exclusively dedicated to language processing, whose precise anatomic location varies across individuals.

To activate the language network, the researchers had children listen to stories inside the MRI scanner. Depending on their age, some heard excerpts of “Alice in Wonderland,” some listened to podcasts and TED talks, and others heard shorter, simpler stories. To watch their brains during a non-linguistic task, the researchers had the children listen to nonsense words.

Across the data from the three labs, which included children between the ages of 4 and 16, as well as adults for comparison, the team saw clear developmental changes in the brain’s response to language. “The integration of the system — how well different subregions of the system correlated with each other and worked together during language processing — was stronger in older children as compared to younger children,” says Ola Ozernov-Palchik, a research scientist in Gabrieli’s lab and a research assistant professor at Boston University. The system was also more strongly activated by language in older children, which may reflect their growing comprehension of what they hear.

But strikingly, almost all language processing happened on the left side of the brain, even in the youngest subjects. “From age 4 on, it seems just as lateralized as in an adult,” Gabrieli says.

Language and developmental disorders

The researchers say this finding has implications for understanding developmental conditions that impact language, including autism and dyslexia. The right side of the brain frequently gets more involved in language processing in people with these conditions than it does in typically developing children. “Almost every single developmental disorder that’s associated with language has a theory that’s related to language lateralization,” Ozernov-Palchik says.

The reason for more bilateral language processing in some disorders is debated. One idea has been that some people might use both sides of their brain for language processing because their brains are less mature. If the right side of the brain processes language early in life, scientists had reasoned, it might simply continue to do so for longer in people with autism or dyslexia than it does in neurotypical individuals. But if most people use the left side of their brains for language even when they are young, the difference can’t be attributed to a developmental delay. Other developmental differences might cause bilateral language processing instead.

The researchers don’t have the full picture yet; they still need to know what parts of the brain process language in children younger than 4. Likewise, they would like to know what the brain areas that become the language network are doing in the first months of life, when infants aren’t using language yet. They are eager to find out, both to understand fundamentals of brain development and to shed light on developmental disorders. “I think understanding that normal trajectory is really critical for interpreting what a deviation from that trajectory is,” says Amanda O’Brien, a former graduate student in Gabrieli’s lab who is now a postdoc at Harvard University.

One reason people thought lateralization might develop gradually is because damage to the left hemisphere of the brain impacts language abilities differently, depending on when it occurs. “If you have damage to the left hemisphere as an adult, you’re very likely to end up with some form of aphasia, at least temporarily,” Fedorenko explains. “But a lot of the time, with early damage to the left hemisphere, you grow up and you’re totally fine. The language can just develop in the right hemisphere.”

Some scientists suspected that the right side of the brain was able to take over language processing in children who suffered early-life brain damage because it was already participating in this function at the time. But the team’s findings suggest the developing brain may be nimbler than that. “Our data tell you that this early plasticity apparently happens in spite of the fact that by age 4, we see these very strongly lateralized responses already,” Fedorenko says.

Researchers “reprogram” materials by quickly rearranging their atoms

Zach Winn | MIT News — Wed, 13 May 2026 11:00:00 -0400

It’s been 37 years since scientists first demonstrated the ability to move single atoms, suggesting the possibility of designing materials atom by atom to customize their properties. Today there are several techniques that allow researchers to move individual atoms in order to give materials exotic quantum properties and improve our understanding of quantum behavior.

But existing techniques can only move atoms across the surface of materials in two dimensions. Most also require painstakingly slow processes and high-vacuum, ultracold lab conditions.

Now a team of researchers at MIT, the Department of Energy’s Oak Ridge National Laboratory, and other institutions has created a way to precisely move tens of thousands of individual atoms within a material in minutes at room temperature. The approach uses a set of algorithms to carefully position an electron beam at specific locations of a material, then scan the beam to drive atomic motions.

“The results demonstrate the ability to deterministically move atoms repeatedly within a material’s 3D atomic lattice,” says MIT Research Scientist Julian Klein, who conceived of and directed the project. “We can reprogram materials to create defects at will, realizing entirely artificial states of matter not found in nature with a wide range of potential applications, including sensing, optical, and magnetic technologies. There are so many opportunities enabled by these techniques.”

“It’s like a photocopier that can create columns of identical atomic defects,” says Frances Ross, MIT’s TDK Professor in Materials Science and Engineering. “It’s especially useful because you can move a few atoms to form defects, and do it again and again to build atomic arrangements in three dimensions that have tunable functions in a system that is more robust because the defects exist beneath the surface.”

In a Nature paper appearing today, the researchers described their approach and how they used it to create more than 40,000 quantum defects in a crystalline semiconductor material.

The researchers say the approach offers a new way to study quantum behavior in materials. It could also one day lead to improvements in systems that leverage quantum defects, like quantum computers, dense magnetic memory, atomic-scale logic devices, and more.

Joining Klein and Ross on the paper are Kevin Roccapriore and Andrew Lupini, researchers at Oak Ridge National Laboratory; Mads Weile, a former MIT visiting student; Sergii Grytsiuk, a former Radbound University researcher; Malte Rösner, a professor at Bielefeld University in Germany; Zdenek Sofer, a professor at the University of Chemistry and Technology Prague in the Czeck Republic; Dimitar Pashov, a research associate at King’s College London; and Mark van Schilfgaarde and Swagata Acharya, researchers at the National Laboratory of the Rockies.

Designing matter

In a now-famous 1989 demonstration, IBM researchers used a scanning tunneling microscope to arrange 35 atoms on the surface of a chilled crystal to spell out “IBM.” It was the first time atoms had been precisely positioned, and an important milestone. The approach enabled scientists to engineer specific defects, such as atom-sized vacancies and surface atoms in crystalline materials, leading to major advances in quantum science. But placing those 35 atoms had taken researchers many hours, if not days.

In parallel with those developments, researchers also developed two additional approaches for manipulating atoms in a vacuum, using optical tweezers to trap neutral atoms and oscillating electric fields to trap ions.

While those approaches have enabled remarkable progress, they remain limited to either surfaces or highly controlled experimental systems. Another factor limiting the design of materials for applications such as quantum computers is the inability of atomic manipulation techniques to move atoms in three dimensions: The patterns are created on the surface of a material, where they are exposed to the environment and cannot survive outside tightly controlled laboratory settings.

Engineering usable materials with custom quantum properties would require researchers to rearrange many more atoms, preferably on the interior of materials. The MIT researchers demonstrated that capability in their Nature study.

“We were trying to improve the number of atoms we could move in a reasonable length of time,” Ross explains. “You want to place the atoms close to each other so they can interact, and you want to have a lot of them arranged as you’d like — thousands or millions of atoms in specific locations you’ve chosen. That’s been challenging with existing techniques.”

The researchers used high-performance microscopes at the Department of Energy’s Oak Ridge National Laboratory for their work. Their new technique uses a sophisticated set of algorithms to direct an electron beam at a target atom with a precision of a few picometers (one trillionth of a meter). The beam does a tight loop to help zero in on its target, then sends a beam of electrons through the material in a carefully designed oscillating path, spending about a second at each location.

“We developed algorithms that allow us to quickly obtain information on where the beam is in the material,” Klein explains. “The trick is to use very few electrons in the process of getting that information, so the whole process is fast and does not unintentionally damage your crystal. It took many years to develop these algorithms and determine the minimum required information needed to infer where the atoms are located with the highest precision.”

The motion of the beam as it delivers electrons, an oscillating path devised by the researchers, pushes entire columns of atoms to new locations the way you might swipe a screen on your phone.

In their experiments, the researchers used this approach to direct the movement of columns of chromium atoms in a stable semiconductor material, chromium sulfide bromide, using a crystal about 13 nanometers thick. The beam created atom-sized vacancies in the material, each vacancy paired with the displaced atom, that they calculated would give the crystal exotic quantum properties.

To show how well their approach scaled, the researchers created over 40,000 defects in about 40 minutes, creating vacancies and interstitials across different distances and in different patterns, calculating that different atomic arrangements should give rise to different quantum mechanical properties.

“Each of these defects has certain ways to interact with its neighbors,” Ross says. “If you place them in a pattern, you could essentially simulate the interactions between the electrons within a molecule, so the whole electronic structure of that molecule can, in a sense, be mapped onto a pattern that you can write into a solid material.”

Probing quantum systems

The success of the approach was likely aided by the way chromium binds within the semiconductor, which has a unique electronic structure. The researchers are further investigating other crystals in which this might work, though they suspect it will be applicable to a diverse range of materials.

In the materials where it works, the approach has several advantages over existing techniques.

“Moving atoms within solids enables the creation of quantum properties in materials that are stable in the air outside of vacuum conditions,” Klein explains. “And this approach is also scalable to many atomic manipulations, so moving thousands or millions of atoms to create artificial structures would represent completely new physics. We’d like to study those systems.”

The researchers say their technique lays the foundation for a new class of programable matter, which could aid the development of a range of stable quantum devices.

“This is a way of accessing physical phenomena that involve a lot of atoms placed in a certain specified arrangement, and can’t be done by self-assembly,” Ross says. “You can create individually tuned atomic arrangements, and you can have so many of them, each arranged exactly how you like over areas that are tens and hundreds of nanometers. That leads to collective physics we are excited to explore.”

The work was supported, in part, by the Department of Energy and the National Science Foundation.

A new approach to cancer vaccination yields more powerful T cells

Anne Trafton | MIT News — Wed, 13 May 2026 05:00:00 -0400

MIT engineers have developed a new way to amplify the T-cell response to mRNA vaccines — an advance that could lead to much more powerful cancer vaccines and stronger protection against infectious diseases.

Most vaccines generate both antibodies and T cells that can target the vaccine antigen by activating antigen-presenting cells, such as dendritic cells. In this study, the researchers boosted the T-cell response with a new type of vaccine adjuvant (a material that can help stimulate the immune system). The new adjuvant consists of mRNA molecules encoding genes that turn on immune signaling pathways and promote a supercharged T-cell response.

In studies in mice, this mRNA-encoded adjuvant enabled the immune system to completely eradicate most tumors, either on its own or delivered along with a tumor antigen. The adjuvant also boosted the T-cell response to vaccines against influenza and Covid-19.

“When these adjuvant mRNAs are included in the vaccines, the number of antigen-targeted T cells is substantially increased. These T cells play an important role in the immune response, assisting in the clearance of virally infected cells or, in the case of cancer, killing cancerous cells,” says Daniel Anderson, a professor in MIT’s Department of Chemical Engineering and a member of MIT’s Koch Institute for Integrative Cancer Research and Institute for Medical Engineering and Science.

Anderson and Christopher Garris, an assistant professor at Harvard Medical School and Massachusetts General Hospital, are the senior authors of the study, which appears today in Nature Biotechnology. The paper’s lead authors are Akash Gupta, a former Koch Institute research scientist who is now an assistant professor at the University of Houston; Kaelan Reed, an MIT graduate student; and Riddha Das, a research fellow at Harvard Medical School and MGH. Robert Langer, the David H. Koch Institute Professor at MIT, and Ralph Weissleder, a professor of radiology and systems biology at MGH and Harvard Medical School, are also authors.

More powerful vaccines

Vaccines that stimulate the body’s immune system to attack tumors have shown promise in clinical trials, and a handful have been FDA-approved for certain cancers. In some patients, these vaccines stimulate a strong response, but in others, a weak response fails to kill the cancerous cells.

The MIT-MGH team wanted to find a way to make those immune responses more powerful. One way to do that is to deliver immune-stimulating molecules called cytokines along with a vaccine. However, cytokines can overstimulate the immune system, leading to potentially severe side effects.

As an alternative approach, the researchers decided to deliver mRNA strands encoding two genes, IRF8 and NIK, which are involved in antigen presentation and can switch immune cells into a more active state.

NIK is an enzyme that activates a signaling pathway involved in immunity and inflammation, while IRF8 is a transcription factor that helps program dendritic cells, particularly a subset called cDC1, which are especially effective at activating T cells. These antigen-presenting cells can digest foreign antigens and present them to T cells, stimulating the T cells to mount an immune response against the antigen.

“We see that the dendritic cells start shifting toward a more cDC1 phenotype, which is the most important dendritic cell phenotype and can generate a stronger T-cell response,” Gupta says.

The researchers packaged the mRNA in lipid nanoparticles similar to those used to deliver mRNA Covid vaccines, but with a different chemical composition that promotes their delivery to the spleen after being injected intravenously.

Inside the spleen, the particles encounter antigen-presenting cells, including dendritic cells. Within 24 hours, these cells begin expressing IRF8 and NIK, and both of these pathways help drive dendritic cells to mature and become activated so that they can prime an anti-tumor response.

Over a few days to a week, the T-cell population expands. These T cells, along with other immune cells such as natural killer (NK) cells, can then recognize and attack tumors.

“Most cancer immunotherapies rely on external signals to activate immune cells. We take a different approach — reprogramming immune cells from within by targeting their internal signaling machinery, enabling a more potent and durable anti-tumor response,” Das says.

Stronger T cells

The researchers tested the immune-remodeling mRNAs in several mouse models of cancer, including an aggressive bladder cancer, colon carcinoma, melanoma, and metastatic lung cancer. In nearly all of these mice, the injected mRNA stimulated a strong T-cell response that significantly slowed tumor growth and in many cases completely eradicated the tumors. This happened even when the mice were not given a vaccine against a specific cancer antigen. When they were, the response was even stronger.

“We showed that you can get an anti-cancer response with these adjuvants without including the antigen, just by activating the immune system. However, cancer-specific antigens with the adjuvants in a vaccine further improved the responses,” Anderson says.

The mRNA adjuvant also enhanced the immune response to immunotherapy drugs called checkpoint blockade inhibitors. These drugs, which work by lifting a brake that tumor cells put on T cells, are FDA-approved to treat several kinds of cancer. These drugs don’t work for all patients, but combining them with the mRNA vaccine adjuvant could offer a way to make them more effective, the researchers say.

“The microenvironment of solid tumors is often hostile to T cells and represents a major barrier to effective immunotherapy. We find that immune remodeling with these adjuvants creates a T cell–permissive environment and promotes tumor rejection,” Garris says.

The researchers also explored whether their new adjuvant could boost the immune response to vaccination against viral infection. When they delivered the mRNA particles along with Covid or flu vaccines, they found that the vaccine generated a 10-to-15-fold stronger T cell response in the mice.

The researchers now plan to test this approach in additional animal models, in hopes of developing it for use in both cancer and infectious diseases.

“While there are differences between the mouse systems that we’ve worked in and humans, we are optimistic that these adjuvants will work in humans and could improve a range of different vaccines,” Anderson says.

The research was funded by Sanofi, the National Institutes of Health, the Marble Center for Cancer Nanomedicine, and the Koch Institute Support (core) Grant from the National Cancer Institute.

A new way to spot signs of dark matter

Jennifer Chu | MIT News — Tue, 12 May 2026 13:00:00 -0400

Dark matter is thought to make up most of the matter in the universe, but the only way it interacts with its surroundings is through gravity. If two colliding black holes spiral through a dense region of dark matter and merge, gravitational waves rippling across space and time could carry an imprint of that dark matter.

Now, physicists may be able to spot such imprints of dark matter in gravitational waves that are detected on Earth.

Researchers at MIT and in Europe have developed a method that makes predictions for what a gravitational wave should look like if it were produced by black holes that moved through dark matter, rather than empty space. They applied the technique to publicly available gravitational-wave data previously recorded by LIGO-Virgo-KAGRA (LVK), the global network of observatories that detect gravitational waves from black hole mergers and other far-off astrophysical sources.

The researchers looked through the gravitational-wave signals recorded over the LVK’s first three observing runs. From 28 of the clearest signals, the team found that 27 originated from black holes that merged in a vacuum, as physicists expected. But the pattern of one signal, GW190728, showed possible signs of a dark matter imprint.

The scientists emphasize that they have not detected dark matter. Rather, the new method offers a new way to screen gravitational-wave data for hints of dark matter, which physicists can then follow up and confirm with other techniques.

“We know that dark matter is around us. It just has to be dense enough for us to see its effects,” says Josu Aurrekoetxea, a postdoc in the MIT Department of Physics. “Black holes provide a mechanism to enhance this density, which we can now search for by analyzing the gravitational waves emitted when they merge.”

Aurrekoetxea and his colleagues report their results in a study appearing today in Physical Review Letters. The study’s co-authors are LVK member Soumen Roy of Université Catholique de Louvain (UCLouvain) in Belgium, Rodrigo Vicente of the University of Amsterdam, Katy Clough of Queen Mary University of London, and Pedro Ferreira of Oxford University.

A dark pull

Dark matter is an invisible, hypothetical form of matter that, unlike normal everyday matter, has no interactions with the electromagnetic force. Dark matter can pass through light, magnetic fields, and any other form of energy along the electromagnetic spectrum without leaving a trace. The only evidence that dark matter exists is through its apparent interaction with one other force: gravity.

By observing how gravity bends around distant galaxies, astronomers have surmised that there must be an extra force, outside of the galaxies’ own gravitational pull, to explain the bending fields, or “lensing.” This extra force, physicists suspect, is dark matter, which could account for over 85 percent of the matter in the universe. But exactly what dark matter is is a matter of huge debate, with theories for dark matter particles that range widely in particle size and properties.

One class of proposed dark matter consists of “light scalar” particles, whose masses are many orders of magnitude lighter than an electron. Theorists predict that such dark matter should behave not just as particles, but also as coordinated waves when moving near black holes.

When waves of dark matter come in contact with a rapidly spinning black hole, physicists predict that the black hole's rotational energy can be transferred to the dark matter, amplifying it. This phenomenon, known as superradiance, would whip up the waves to extremely high densities of dark matter, akin to churning cream into butter.

At high enough densities, light scalar dark matter, which is invisible by all other accounts, should leave an imprint on the gravitational waves that reverberate from the colliding black holes.

But exactly what would that imprint look like? And could such an imprint be detectable in gravitational waves that arrive on Earth, from black holes that merged many millions of light years away?

For answers to those questions, Aurrekoetxea and his colleagues developed a model to predict the gravitational waveform, or the pattern of gravitational waves that two black holes would produce, if they collided in an environment of dark matter, versus in a vacuum (empty space, with no dark matter).

An imprint’s prediction

For their new study, the team performed detailed numerical simulations to predict the gravitational wave that would be produced given various properties of two colliding black holes — a system known as a “black hole binary.” They considered black hole binaries across a range of scenarios and properties, for example, varying the size and mass of each black hole, the environment of dark matter that the black holes might pass through, and the density of the dark matter that the black holes would spin up.

They designed the model to predict what a gravitational wave from a black hole binary would look like if it carried an imprint of dark matter, and furthermore, what that wave would look like if it traveled a given distance across space and time, to eventually arrive at a detector on Earth.

With their model, they looked to see whether any gravitational-wave signals that have been detected on Earth match their predicted patterns of dark matter imprints. To do so, they applied the model to publicly-available data recorded by LVK over the observatories’ first three observing runs. The observatories have picked up hundreds of gravitational-wave signals during this period. For their purposes, the researchers focused on the clearest signals, comprising gravitational waves from 28 separate events.

For each event, the team compared the pattern of the actual gravitational wave against their model of what the signal would look like if it were generated by the same event in an environment of dark matter. They also compared the gravitational wave to the more expected scenario in which the signal was produced in a vacuum.

Of the 28 clearest signals that they analyzed, 27 were solidly within the predictions for having been produced in a vacuum. However, the pattern of one event, GW190728, showed a “preference,” or an agreement with the team’s dark matter model. In other words, the signal may carry an imprint of dark matter.

GW190728 is a gravitational wave that is named after the date that it was detected — on July 28, 2019. Scientists previously determined that the gravitational wave originated from a black hole binary with a total mass of about 20 times the mass of the sun. With their model, the team showed that such a system could have merged through a dense cloud of dark matter and produced a similar gravitational wave to GW190728.

“The statistical significance of this is not high enough to claim a detection of dark matter, and further checks should be performed by independent groups,” Aurrekoetxea says. “What we think is important to highlight is that without waveform models like ours, we could be detecting black hole mergers in dark matter environments, but systematically classifying them as having occurred in vacuum.”

“We now have the potential to discover dark matter around black holes as the LVK detectors keep collecting data in the coming years,” says co-author Soumen Roy, who led the data analysis part of the work. “It is an exciting time to search for new physics using gravitational waves.”

“Using black holes to look for dark matter would be fantastic,” adds co-author Rodrigo Vicente, who developed the analytical model of the signal. “We would be able to probe dark matter at scales much smaller than ever before.”

This work was supported, in part, by the U.S. National Science Foundation and MIT’s Center for Theoretical Physics — a Leinweber Institute.

Powerful shrinking technique could enable devices that compute with light

Anne Trafton | MIT News — Tue, 12 May 2026 05:00:00 -0400

Using a new technique that can create vacancies at any site across a material and then shrink it to about 1/2,000 of its original volume, MIT researchers have designed nanotechnology devices that could be used for optical computing and other applications involving the manipulation of visible light.

The new fabrication technique, known as “implosion carving,” allows researchers to imprint features throughout a hydrogel using photopatterning. If patterned with a resolution of about 800 nanometers, these features can then be shrunk to less than 100 nanometers.

Because that resolution is smaller than the wavelength of light, the devices can bend light in specific ways that allow them to perform optical computations.

“In order to enable nanophotonic applications in visible light, we need to make nanostructures with feature sizes with a resolution less than 100 nanometers. Only in that way can we precisely create the structure that can manipulate visible light,” says Quansan Yang, a former MIT postdoc, now an assistant professor at the University of Washington, and one of the lead authors of the new study.

In their paper, the researchers demonstrated a photonic device that can perform a simple digit-classification task, but future versions could be used for high-speed imaging and information processing, they say.

Gaojie Yang, a former MIT postdoc, is the co-lead author of the paper, which appears today in Nature Photonics. The paper’s senior authors are Peter So, director of the MIT Laser Biomedical Research Center (LBCR) and an MIT professor of biological engineering and mechanical engineering, and Edward Boyden, the Y. Eva Tan Professor in Neurotechnology at MIT and a professor of biological engineering, media arts and sciences, and brain and cognitive sciences. Boyden is also a Howard Hughes Medical Institute investigator and a member of MIT’s McGovern Institute for Brain Research, the Yang Tan Collective, and Koch Institute for Integrative Cancer Research.

Nanoscale feature sizes

Photonic devices, which transmit and manipulate light, hold potential for use as optical computer chips that could offer an energy-efficient alternative to semiconductor chips. However, existing techniques for creating 3D photonic devices haven’t yet achieved the 100-nanometer resolution that is needed to channel visible light, which has wavelengths between 380 and 750 nanometers.

Using an additive manufacturing technique called two-photon lithography, researchers can use light to create 3D nanoscale features, but with a resolution larger than 100 nanometers. Another technique, known as electron-beam lithography, can be used to etch smaller-resolution features onto a silicon chip, but it doesn’t generate 3D structures.

To make 3D devices with the necessary feature size, the researchers extended the concept of “implosion fabrication,” which Boyden’s lab developed in 2018, to create a new variant called “implosion carving.” In implosion carving, a laser creates vacancies — tiny voids where the hydrogel material has been removed — at precisely targeted locations. These vacancies exhibit different optical properties than the surrounding hydrogel. The hydrogel is then shrunk to bring the patterned features down to the nanoscale.

The carving process begins with immersing the hydrogel in a photosensitizing dye. Then, the researchers use a laser to excite the photosensitizer at specific places in the gel, which in turn generates reactive oxygen species that cut the bonds holding the hydrogel together. This creates a vacancy in that spot.

Once the desired vacancy pattern has been carved into the hydrogel, the researchers shrink it using a two-step process. First, they soak it in a solution containing ions, which causes it to shrink about tenfold in each dimension. To shrink it a little more, and to remove the watery solution, the hydrogel then undergoes a process called supercritical drying, which can remove liquid from a gel without damaging it.

At the end of the process, the hydrogel has been shrunk more than tenfold in each dimension, leading to a 2,000-fold reduction in volume.

Computing with light

To demonstrate the versatility of this technique, the researchers used it to create several 3D shapes, including a helix and a structure inspired by a butterfly wing. Some of these structures are too thin, and have too high an aspect ratio, to be stably created using conventional two-photon lithography.

The researchers also created a device that could perform a simple calculation known as digit classification, a task that is traditionally used to test the performance of neural networks. During this task, the device was presented with a digit, such as 1 or 5, and had to light up a specific location to indicate which number was detected.

To achieve this, the researchers patterned vacancies throughout the device so that it would act like a neural network. The pattern of vacancies would diffract input light as it passed through many layers of patterned hydrogel, so that the output light was determined by the shape of the digit that was entered into the system.

“This is a purely optical system that effectively performs optical computing,” So says.

“One of the very attractive features of this technology is that you can manipulate the property of the material at every tiny location,” says Dushan Wadduwage, an assistant professor at Old Dominion University and former MIT postdoc, who is also an author of the paper. “You have millions of different locations that you need to decide the property of, and that turns into a really interesting design problem where we can use deep-learning algorithms to find designs over these millions of parameters and come up with parts that go into optical systems in new ways.”

The researchers now plan to use the same principles to build optical devices that could classify cells based on their state as they flow through a microfluidic device. This could help identify rare cells such as circulating tumor cells in a blood sample, they say.

This approach could also enable the creation of high-throughput imaging techniques for applications such as analyzing tissue samples from biopsies or surgical specimens. And, if adapted to work with other materials such as hydrophobic polymers, it could also be used to create channels within 3D nanofluidic devices.

Other authors of the paper include Gaojie Yang, Takahiro Nambara, Hiroyuki Kusaka, Yuichiro Kunai, Alex Matlock, Corban Swain, Brett Pryor, Yannick Salamin, Daniel Oran, Hasindu Kariyawasam, Ramith Hettiarachchi, and Marin Soljacic.

The research was funded, in part, by the MIT-Fujikura Partnership Fund, the U.S. Army Research Office through the Institute for Soldier Nanotechnologies at MIT, Lisa Yang and Y. Eva Tan, John Doerr, the Open Philanthropy Project, the Howard Hughes Medical Institute, and the U.S. National Institutes of Health.

Improving the reliability of circuits for quantum computers

Adam Zewe | MIT News — Tue, 12 May 2026 05:00:00 -0400

Quantum computers could someday solve pressing problems that are too convoluted for classical computers, such as modeling complex molecular interactions to streamline drug discovery and materials development.

But to build a superconducting quantum computer that is large and resilient enough for real-world applications, scientists must precisely engineer thousands of quantum circuits so they perform operations with the lowest possible error rate.

To help scientists design more predictable circuits, researchers from MIT and Lincoln Laboratory developed a technique to measure a property that can unexpectedly cause a superconducting quantum circuit to deviate from its expected behavior. Their analysis revealed the source of these distortions, known as second-order harmonic corrections, leading to underperforming circuit architectures.

The MIT researchers fabricated a device to detect second-order harmonic corrections, identify their origin, and precisely measure their strength. This technique could help scientists deliberately design quantum circuits that can counteract the effects of these deviations.

This is especially important in larger and more complicated quantum circuits, where the negative impact of second-order harmonic corrections can be amplified.

“As we make our quantum computers bigger and we want to have more precise control over the parameters of these devices, identifying and measuring these effects is going to be important for us to have a precise understanding of how these systems are constructed. It is always important to keep diving down into the circuit to see if there is an effect you didn’t expect, which impacts how your device is performing,” says Max Hays, a research scientist in the Engineering Quantum Systems (EQuS) group of the Research Laboratory of Electronics (RLE) and co-lead author of a paper on this research.

Hays is joined on the paper by co-lead author Junghyun Kim, an electrical engineering and computer science (EECS) graduate student in the EQuS group; senior author William D. Oliver, the Henry Ellis Warren (1894) Professor of EECS and professor of physics, leader of the EQuS group, director of the Center for Quantum Engineering, and associate director of RLE; as well as others at MIT and Lincoln Laboratory. The research appears today in Nature Physics.

A pair-wise problem

In a quantum computer that utilizes superconducting circuits, which is one of many potential computing platforms, Josephson junctions are critical elements that enable the transfer and manipulation of information. These devices utilize two superconducting wires that are brought very close together, with a nanometer-scale barrier between them. Like a traditional circuit, the electric charge in Josephson junctions is carried by electrons.

But in a superconducting circuit, charge-carrying electrons pair up, forming what are called Cooper pairs. These Cooper pairs can “quantum tunnel” through the barrier between the two wires, transporting current from one wire to the other.

Cooper pairs can usually only tunnel one pair at a time, which is a key property that makes quantum computation possible.

“If you try to force more Cooper pairs through, it just doesn’t work. This non-linear effect is extremely important for all our circuits. If we didn’t have that effect, then we wouldn’t be able to control or manipulate any quantum information that we store in these circuits,” Hays explains.

But sometimes, Cooper pairs can unexpectedly squeeze through the barrier two at a time, an effect that is known as a second-order harmonic correction. This effect limits the performance of a quantum circuit that has been configured to only allow single-pair tunneling.

“If two Cooper pairs tunnel at the same time, then the assumption we used to build our circuit doesn’t apply anymore. We need to fix the circuit so it can handle that,” Kim says.

But before they can fix the circuit, scientists need to know the source and strength of these distortions.

To obtain this information, the MIT researchers fabricated a quantum circuit so it would be very sensitive to these effects. Essentially, the device is designed to suppress the quantum tunneling process of single Cooper pairs, while allowing the two-pair tunneling process to continue.

In this way, they can detect the presence of second-order harmonic corrections and precisely measure their strength.

Straight to the source

They can also use this circuit to pinpoint the source of these harmonics, which helps researchers identify the best way to correct for them.

There are two potential sources of second-order harmonics — one source is intrinsic to the dynamics of the Josephson junction and the other is caused by the wires connecting the junction to other circuit elements.

While prior research had indicated the second-order harmonics could be due to the dynamics of the junction, the MIT researchers found that additional inductance — the tendency to oppose changes in the flow of electric current —from wires in the circuit was the actual source in their devices.

“This is important because, if we know where the second-order harmonic correction is coming from, we can predict how strong it is likely to be, and use that information to engineer more predictable circuits that will hopefully perform better,” Hays says.

In the future, the researchers want to design experiments that more accurately predict how a device will perform when second-order harmonic corrections occur. They also want to study other sources of second-order harmonic corrections and whether those sources could have negative impacts on a circuit under different fabrication conditions.

This work is funded, in part, by the U.S. Department of Energy, the U.S. Co-design Center for Quantum Advantage, the U.S. Air Force, the Korea Foundation for Advanced Studies, and the Intelligence Community Postdoctoral Research Fellowship Program at MIT.

For most US drivers, EVs offer emissions benefits and cost savings

Adam Zewe | MIT News — Tue, 12 May 2026 00:00:00 -0400

Despite regional variability in climate, electricity sources, congestion, and the wide variation in individual driving patterns, electric vehicles generate less greenhouse gas emissions and do not cost more than comparable gas-powered vehicles for drivers and vehicle fleet owners in most parts of the United States, according to a new study by MIT researchers.

The team’s approach captures many key factors that contribute to regional and individual differences in the life-cycle emissions and ownership cost of electric vehicles, including meteorological data, the distance and duration of trips, and fuel prices.

To paint a fuller picture of emissions and costs than was previously available, the researchers sourced data from thousands of U.S. zip codes and drilled down to the level of individual drivers within those locations. Their study considers time-averaged fuel prices so as not to be overly influenced by fluctuations in prices at any one point in time. They finalized their analysis at the end of 2024 and early 2025.

Their results indicate that a person’s driving behaviors can matter as much as regional factors like the local electricity mix when it comes to the emissions savings of an electric vehicle, compared to a similar gas-powered vehicle. In most locations, a battery-electric vehicle reduces emissions between 40 and 60 percent, with larger impacts in urban areas.

They also found that colder climates do not reduce overall emission benefits as much as some media reports assume.

The researchers utilized this detailed analysis to update a public tool they previously developed, carboncounter.com, which enables individuals to compare the life-cycle emissions and total ownership costs of nearly any car on the market. A new version of carboncounter.com is also being released today.

“There are a lot of statements being thrown around, like that electric vehicles don’t reduce emissions very much in cool climates, and we wanted to analyze these factors systematically and evaluate these statements against one another simultaneously. Rather than simply asking, ‘Are EVs better?’, this paper helps answer ‘better for whom, and under what conditions?’” says Marco Miotti PhD ’20, a senior researcher at ETH Zurich who completed this research while a graduate student in the Institute for Data, Systems, and Society (IDSS) at MIT.

He is joined on the paper by senior author Jessika Trancik, a professor in IDSS. The research appears today in Environmental Research Letters.

A holistic approach

Many prior studies that compare emissions and costs of electric vehicles (EVs) to combustion-engine vehicles cover a few factors, like the amount of renewable energy in the grid and how gas prices impact affordability, Miotti says.

“To our knowledge, there have been few efforts so far that bring all these factors together. But if someone wants to buy a car and have a better understanding of the factors that affect emissions and costs, this holistic approach is important,” he adds.

The researchers focused on two types of EVs: battery-electric vehicles, which only operate on electricity, and plug-in hybrid electric vehicles, which also have a combustion engine that works in tandem with the battery to optimize fuel savings.

The team expanded and improved a set of previously developed vehicle cost and emissions models to incorporate a wider variety of factors and data types.

For instance, they refined an existing model that estimates energy use and gas mileage so it could capture more nuances of local climate variability.

“But the real effort was not just in extending these different models, but in bringing together all these different data and making them work with the models in a consistent manner,” Miotti says.

The team sourced data on a wide variety of factors for each U.S. zip code, such as typical drive cycles, the amount of traffic, local gas and electricity prices, makeup of the regional electricity mix, meteorological profiles, and more. They used statistical approaches to amalgamate different types of data.

For example, the team used a probabilistic matching technique to combine data on how often people drive, which was drawn from nationwide travel surveys, with more detailed GPS data that includes factors like drivers’ acceleration patterns and the distance they usually drive on each day of the week.

The researchers designed their analysis to focus on the spatial picture of emissions and costs, based on U.S. zip codes, while simultaneously considering the impact of the size and features of each specific vehicle model.

“At the end of the day, it’s the vehicle and fleet owners who make decisions about vehicle purchases. So, we wanted to make sure to consider their wide-ranging individual perspectives rather than simply performing a region-by-region comparison,” says Trancik.

Lower emissions, comparable costs

In the end, their modeling framework revealed that all factors they analyzed matter about equally in determining emissions-reduction potential of EVs compared to internal combustion vehicles.

EVs reduce emissions the most in areas with a cleaner electricity mix, denser traffic, higher annual travel distances, and a mild climate, in decreasing order of importance. In each area, emission reductions increase for drivers who drive more often, drive larger vehicles, and are more frequently stuck in traffic.

In a colder area like North Dakota, fuel economy of battery-electric vehicles might be reduced by as much as 50 percent on a particularly frigid night, but the effect on annual emission benefits is minimal.

“We even did a sensitivity study to see if the range is reduced in very cold climates, and we found that, even in the most unfavorable conditions, EVs still reduce emissions by a substantial amount,” Miotti says.

On the cost side, the models show that, in most places across the U.S., EVs are competitive with comparable combustion-engine vehicles in terms of lifetime ownership cost, even without clean vehicle tax credits. And in areas where electricity is relatively affordable, battery-electric vehicles tend to cost less than their plug-in hybrid or combustion-engine counterparts.

In the future, the researchers want to expand this analysis to include a temporal dimension, so the framework also considers how changes in vehicle, fuel, and electricity prices affect emissions and costs over time.

“While we found that the electricity mix is a big driver of the spatial variation in emissions savings of EVs, the electricity grid is decarbonizing everywhere. As that happens, emissions savings across space will become more homogenous for EVs, but the differences across one driver to another will remain,” Miotti says.

They could also use the framework to explore regions outside the United States or incorporate data on hybrid-electric vehicles that cannot be plugged in.

This work was funded, in part, by the MIT Martin Family Society of Fellows for Sustainability.

Photonics advance could enable compact, high-performance lidar sensors

Adam Zewe | MIT News — Thu, 07 May 2026 05:00:00 -0400

Lidar systems use pulses of infrared light to measure distance and map a 3D scene with high resolution, allowing autonomous vehicles to rapidly react to obstacles that appear in their path. But traditional lidar sensors are expensive, bulky systems with many moving parts that degrade over time, limiting how the sensors can be deployed.

A new study from MIT researchers could help to enable next-generation lidar sensors that are compact, durable, and have no moving parts. The key advance is a novel design for a silicon-photonics chip, which is a semiconductor device that manipulates light rather than electricity.

Typically, such silicon-photonics chip-based systems have a restricted field of view, so a silicon-photonics-based lidar would not be able to scan angles in the periphery. Existing workarounds to this problem increase noise and hamper precision.

To avoid these drawbacks, the MIT researchers designed and demonstrated an array of integrated antennas that minimizes unwanted crosstalk between the antennas. Their innovation allows a lidar chip to scan a wider field of view while maintaining low-noise operation compared to other silicon-photonics-based approaches.

This novel demonstration could fuel the development of advanced lidar sensors for demanding applications like autonomous vehicle navigation, aerial surveying, and construction site monitoring.

“The functionality we demonstrated in this work solves a fundamental problem for integrated optical-phased-array technology, enabling future lidar sensors that can achieve significantly higher performance than we could demonstrate previously,” says Jelena Notaros, the Robert J. Shillman Career Development Associate Professor of Electrical Engineering and Computer Science (EECS) at MIT, a member of the Research Laboratory of Electronics, and senior author of a paper on this innovation.

She is joined on the paper by lead author and EECS graduate student Henry Crawford-Eng as well as EECS graduate students Andres Garcia Coleto, Benjamin M. Mazur, Daniel M. DeSantis, and Tal Sneh. The research appears today in Nature Communications.

Adjusting an antenna array

Many traditional lidar systems map a scene using a bulky box that spins to send pulses of light in multiple directions. The light bounces off nearby objects and returns to the sensor, providing data that are used to reconstruct the environment.

Instead, silicon-photonics-based lidar sensors systematically scan an emitted light beam in multiple directions non-mechanically using a system called an integrated optical phased array (OPA).

Key to an OPA is an array of integrated antennas that have tiny perturbations placed periodically along their length. These corrugations allow the antenna to scatter light from an input source up and out of the photonic chip.

By adjusting the phase of light routed to each antenna, the researchers can change the angle at which the light is emitted out of the array. In this way, they can steer the beam with no moving parts.

But if engineers place the antennas too close together, the antennas will couple with each other and the light they emit will get jumbled. To avoid this, scientists typically space the antennas farther apart, but this also has downsides.

If the antennas are spaced too far apart, the array will emit multiple copies of the light beam at different angles. The researchers can only steer the primary beam so far in either direction until it is undiscernible from its neighboring copies.

“This limits our field of view, so the autonomous vehicle now only knows what is in front of it for a certain angular range,” Garcia Coleto explains.

These beam copies, known as grating lobes, can cause false positives by confusing the sensor. They also waste power.

The MIT researchers solved this problem by designing a set of reduced-crosstalk antennas that can be placed close together without causing a significant coupling effect.

In a standard OPA, all the antennas have the same design, meaning the same arrangement of corrugations. These identical antennas couple very strongly when placed close together.

To address this fundamental roadblock, the MIT researchers designed a set of three antennas with different geometries, varying the width of each antenna and the size and arrangement of corrugations. With varied geometries, each antenna has a different propagation coefficient, which determines how light travels down the antenna.

“Because the antennas have very different propagation coefficients, when we put them close together, essentially each antenna doesn’t ‘see’ the antenna next to it. Therefore, it won’t couple with its neighbor,” Garcia Coleto says.

A photonic balancing act

But even though the antennas have different propagation coefficients, the researchers still need them to emit light in the same way.

They achieved this by carefully designing the antennas to meet three parameters.

First, each antenna must emit the same amount of light. Second, each antenna must emit a beam at the same angle for the same wavelength of light. Third, the emission angle must change uniformly across the array as the researchers steer it.

“We have this challenge where we require the antennas to have different geometries to reduce the crosstalk, but we need to simultaneously design the antennas to have the same emission characteristics. While it is possible to engineer this, it is extremely difficult because, typically, when antennas are designed with different geometries, they tend to behave differently,” Crawford-Eng says.

The researchers first developed the fundamental electromagnetic theory behind how radiative modes couple. They used that theory as a guide to design and simulate their antennas.

Building on those analyses, they fabricated the OPA with reduced-crosstalk antennas spaced significantly closer than they would be in a traditional OPA, then experimentally tested the system.

While a typical OPA would have coupling of about 100 percent in this experiment, their OPA reduced coupling to about 1 percent while generating a single, precise beam. Using this design, they demonstrated accurate beam steering across a wide field of view without any grating lobes.

In the future, the researchers plan to further improve their technique to enable an even wider field of view. In addition, they are exploring a new potential solution to wide field-of-view functionality that they discovered while developing the underlying theory.

“This work addresses a longstanding challenge in integrated optical phased arrays: simultaneously achieving both a wide field of view, which requires dense antenna spacing, and high beam quality, which requires low crosstalk between neighboring antennas. The authors solve this problem with an elegant antenna design. Their innovation is an important step forward for chip-scale, solid-state beam-steering technology,” says Joyce Poon, professor of electrical and computer engineering at the University of Toronto and director of the Max Planck Institute of Microstructure Physics, who was not involved with this work.

This research was supported, in part, by the Semiconductor Research Corporation, the National Science Foundation, an MIT MathWorks Fellowship, the U.S. Department of War, and the MIT Rolf G. Locher Endowed Fellowship.

The work was conducted, in part, using MIT.nano facilities.

Study: Firms often use automation to control certain workers’ wages

Peter Dizikes | MIT News — Thu, 07 May 2026 00:00:00 -0400

When we hear about automation and artificial intelligence replacing jobs, it may seem like a tsunami of technology is going to wipe out workers broadly, in the name of greater efficiency. But a study co-authored by an MIT economist shows markedly different dynamics in the U.S. since 1980.

Rather than implement automation in pursuit of maximal productivity, firms have often used automation to replace employees who specifically receive a “wage premium,” earning higher salaries than other comparable workers. In practice, that means automation has frequently reduced the earnings of non-college-educated workers who had obtained better salaries than most employees with similar qualifications.

This finding has at least two big implications. For one thing, automation has affected the growth in U.S. income inequality even more than many observers realize. At the same time, automation has yielded a mediocre productivity boost, plausibly due to the focus of firms on controlling wages rather than finding more tech-driven ways to enhance efficiency and long-term growth.

“There has been an inefficient targeting of automation,” says MIT’s Daron Acemoglu, co-author of a published paper detailing the study’s results. “The higher the wage of the worker in a particular industry or occupation or task, the more attractive automation becomes to firms.” In theory, he notes, firms could automate efficiently. But they have not, by emphasizing it as a tool for shedding salaries, which helps their own internal short-term numbers without building an optimal path for growth.

The study estimates that automation is responsible for 52 percent of the growth in income inequality from 1980 to 2016, and that about 10 percentage points derive specifically from firms replacing workers who had been earning a wage premium. This inefficient targeting of certain employees has offset 60-90 percent of the productivity gains from automation during the time period.

“It’s one of the possible reasons productivity improvements have been relatively muted in the U.S., despite the fact that we’ve had an amazing number of new patents, and an amazing number of new technologies,” Acemoglu says. “Then you look at the productivity statistics, and they are fairly pitiful.”

The paper, “Automation and Rent Dissipation: Implications for Wages, Inequality, and Productivity,” appears in the May print issue of the Quarterly Journal of Economics. The authors are Acemoglu, who is an Institute Professor at MIT; and Pascual Restrepo, an associate professor of economics at Yale University.

Inequality implications

Dating back to the 2010s, Acemoglu and Restrepo have combined to conduct many studies about automation and its effects on employment, wages, productivity, and firm growth. In general, their findings have suggested that the effects of automation on the workforce after 1980 are more significant than many other scholars have believed.

To conduct the current study, the researchers used data from many sources, including U.S. Census Bureau statistics, data from the bureau’s American Community Survey, industry numbers, and more. Acemoglu and Restrepo analyzed 500 detailed demographic groups, sorted by five levels of education, as well as gender, age, and ethnic background. The study links this information to an analysis of changes in 49 U.S. industries, for a granular look at the way automation affected the workforce.

Ultimately, the analysis allowed the scholars to estimate not just the overall amount of jobs erased due to automation, but how much of that consisted of firms very specifically trying to remove the wage premium accruing to some of their workers.

Among other findings, the study shows that within groups of workers affected by automation, the biggest effects occur for workers in the 70th-95th percentile of the salary range, indicating that higher-earning employees bear much of the brunt of this process.

And as the analysis indicates, about one-fifth of the overall growth in income inequality is attributable to this sole factor.

“I think that is a big number,” says Acemoglu, who shared the 2024 Nobel Prize in economic sciences with his longtime collaborators Simon Johnson of MIT and James Robinson of the University of Chicago.

He adds: “Automation, of course, is an engine of economic growth and we’re going to use it, but it does create very large inequalities between capital and labor, and between different labor groups, and hence it may have been a much bigger contributor to the increase in inequality in the United States over the last several decades.”

The productivity puzzle

The study also illuminates a basic choice for firm managers, but one that gets overlooked. Imagine a type of automation — call-center technology, for instance — that might actually be inefficient for a business. Even so, firm managers have incentive to adopt it, reduce wages, and oversee a less productive business with increased net profits.

Writ large, some version of this seems to have been happening to the U.S. economy since 1980: Greater profitability is not the same as increased productivity.

“Those two things are different,” says Acemoglu. “You can reduce costs while reducing productivity.”

Indeed, the current study by Acemoglu and Restrepo calls to mind an observation by the late MIT economist Robert M. Solow, who in 1987 wrote, “You can see the computer age everywhere but in the productivity statistics.”

In that vein, Acemoglu observes, “If managers can reduce productivity by 1 percent but increase profits, many of them might be happy with that. It depends on their priorities and values. So the other important implication of our paper is that good automation at the margins is being bundled with not-so-good automation.”

To be clear, the study does not necessarily imply that less automation is always better. Certain types of automation can boost productivity and feed a virtuous cycle in which a firm makes more money and hires more workers.

But currently, Acemoglu believes, the complexities of automation are not yet recognized clearly enough. Perhaps seeing the broad historical pattern of U.S. automation, since 1980, will help people better grasp the tradeoffs involved — and not just economists, but firm managers, workers, and technologists.

“The important thing is whether it becomes incorporated into people’s thinking and where we land in terms of the overall holistic assessment of automation, in terms of inequality, productivity and labor market effects,” Acemoglu says. “So we hope this study moves the dial there.”

Or, as he concludes, “We could be missing out on potentially even better productivity gains by calibrating the type and extent of automation more carefully, and in a more productivity-enhancing way. It’s all a choice, 100 percent.”

Method for stress-testing cloud computing algorithms helps avoid network failures

Adam Zewe | MIT News — Wed, 06 May 2026 00:00:00 -0400

Researchers from MIT and elsewhere have developed a more user-friendly and efficient method to help networking engineers identify potential system failures before they cause major problems, like a cloud service outage that leaves millions of users unable to access applications.

The technique uncovers hidden blind spots that might cause a shortcut algorithm to fail unexpectedly when it is deployed.

This new approach can identify worse-case scenarios that an engineer might miss if they use a traditional method that compares an algorithm against a set of human-designed past test cases. It is also less labor-intensive than other verification tools that require engineers to rewrite an algorithm in a complex mathematical code each time they want to test it.

Instead of needing a mathematical reformulation, the new method reads the algorithm’s source code directly and automatically searches for worse-case scenarios that lead to the highest level of underperformance.

By helping engineers quickly and easily stress-test a networking algorithm before deployment, the method could catch failure modes that might otherwise only appear in a real outage. The technique could also be used to analyze the risks of deploying AI-generated code.

“We need to have good tools to measure the worse-case scenario performance of our algorithms so we know what could happen before we put them into production. This is an easy-to-use tool that can be plugged into current systems so we can find the best algorithm to use and ensure the worse-case scenarios are identified in advance,” says Pantea Karimi, an electrical engineering and computer science (EECS) graduate student and lead author of a paper on this new technique.

She is joined on the paper by senior authors Mohammad Alizadeh, an associate professor of EECS and a member of the Computer Science and Artificial Intelligence Laboratory (CSAIL); and Behnaz Arzani, a principal researcher at Microsoft Research; along with Ryan Beckett, Siva Kesava Reddy Karkarla, and Pooria Namyar, researchers at Microsoft Research; and Santiago Segarra, a professor at Rice University. The research will be presented at the USENIX Symposium on Networked Systems Design and Implementation.

Assessing algorithms

In large systems like cloud servers, the tried-and-true algorithms that route data from one place to another or are often too computationally intensive to run in a feasible amount of time.

So, engineers and researchers develop suboptimal algorithms called heuristics that can run much faster. However, there could be unexpected but plausible circumstances that will cause a heuristic to underperform or fail when deployed.

A heuristic can route millions of data requests across a cloud network in seconds, but under the wrong conditions — like an unusual traffic pattern or a sudden spike in demand — the shortcut can break down in ways the designer never anticipated.

When these problems occur, a company may have no choice but to drop some requests that can’t be processed.

The firm could also deliberately allocate more resources in advance to head-off a potential disaster, leading to higher overall costs and wasted electricity from underutilization.

“This is really bad for a company because, either way, they are going to lose a lot of money. If this particular scenario hasn’t happened before and was never tested, how would a developer know in advance before it happens?” Karimi says.

Stress-testing heuristics typically involves running a new algorithm in simulation using a set of human-designed test cases and manually comparing the performance with a previous algorithm. But this is time-consuming and can leave blind spots if an engineer doesn’t know to test for certain situations.

Alternatively, engineers could use a verification tool to evaluate the performance of their heuristic more systematically. However, these tools require the engineer to encode the algorithm into a complex, mathematical formula that can take days to flesh out. The process, which doesn’t work for every type of heuristic, must be repeated each time the engineer changes the code.

Instead, the researchers developed a more user-friendly and efficient verification tool, called MetaEase, that analyzes the heuristic’s existing implementation code directly to identify the biggest risks of deploying it.

“This would reduce the friction of using these heuristic analysis tools,” Karimi says.

She began this work during an internship at Microsoft Research, where the team previously developed MetaOpt, a heuristic analyzer that requires engineers to rewrite their algorithms as formal optimization models. MetaEase grew out of the desire to remove that barrier.

Maximizing the gap

MetaEase is driven by two key innovations. First, it uses a technique called symbolic execution to map out the different decision points in the heuristic's code. These are places where the algorithm might behave differently depending on the input.

This technique produces a set of representative starting points, each corresponding to a distinct behavior the heuristic could exhibit.

Second, from these starting points, MetaEase utilizes a guided search to systematically move toward inputs that make the heuristic perform as poorly as possible, compared to the optimal algorithm.

In machine learning, for instance, an input could be a set of user queries to an AI chatbot at a given time.

“In this way, we have exploited every possible heuristic behavior and used special techniques to move in the direction where we think the performance gap is going to increase,” Karimi explains.

In the end, MetaEase identifies the input that maximizes the performance gap between the heuristic and an optimal benchmark.

With this information, a heuristic developer could inspect the input to understand what went wrong and incorporate safeguards that will prevent the problem from happening during deployment.

In simulated experiments, MetaEase often identified inputs with larger performance gaps than traditional methods — pinpointing more catastrophic worse-case scenarios. And it did so much more efficiently.

It was also able to analyze a recent networking heuristic that no state-of-the-art method could handle.

In the future, the researchers want to enhance MetaEase so it can process additional types of types of data, like categorical inputs. They also want to improve the scalability of their method and adapt MetaEase to evaluate more complex heuristics.

“Reasoning about the worst-case performance of deployed heuristics is a hard and longstanding problem. MetaEase makes tangible progress by analyzing heuristics directly from source code, eliminating the need for formal models that have historically limited who can use such analysis tools. I was pleasantly surprised that it handles non-convex and randomized heuristics by combining symbolic execution with gradient-based search in a practical and effective way,” says Ratul Mahajan of the University of Washington Paul G. Allen School of Computer Science and Engineering, who was not involved with this research.

This research was funded, in part, by a Microsoft Research internship and the U.S. National Science Foundation (NSF).

Astronomers pin down the origins of a planetary odd couple

Jennifer Chu | MIT News — Tue, 05 May 2026 00:00:00 -0400

Across the Milky Way galaxy, a planetary odd couple is circling a star some 190 light years from Earth. A normally “lonely” hot Jupiter is sharing space with a mini-Neptune, in a rare and unlikely pairing that’s had astronomers puzzled since the system’s discovery in 2020.

Now MIT scientists have caught a glimpse into the atmosphere of the mini-Neptune, which is circling inside the orbit of its Jupiter-sized companion, and discovered clues to explain the origins of this unusual planetary system.

In a study appearing today in Astrophysical Journal Letters, the scientists report on new measurements of the mini-Neptune’s atmosphere, made using NASA’s James Webb Space Telescope (JWST). It is the first time astronomers have measured the composition of a mini-Neptune that resides inside the orbit of a hot Jupiter.

Their measurements reveal that the smaller planet has a “heavy” atmosphere that is rich with water vapor, carbon dioxide, sulfur dioxide, and hints of methane. Such a heavy atmosphere would not have been acquired by the planet if it had formed in its current location, very close to its star.

Instead, the scientists say their findings point to an alternate origin story: Both the mini-Neptune and the hot Jupiter may have formed much farther away, in the colder region of the protoplanetary disk. There, the planets could slowly build up atmospheres of ice and other volatiles. Over time, the planets were likely drawn in toward the star in a gradual process that kept them close, with their atmospheres intact.

The team’s results are the first to show that mini-Neptunes can form beyond a star’s “frost line.” This boundary refers to the minimum distance from a star where the temperature is low enough that water instantly condenses into ice.

“This is the first time we’ve observed the atmosphere of a planet that is inside the orbit of a hot Jupiter,” says Saugata Barat, a postdoc in MIT’s Kavli Institute for Astrophysics and Space Research and the lead author of the study. “This measurement tells us this mini-Neptune indeed formed beyond the frost line, giving confirmation that this formation channel does exist.”

The team consists of astronomers around the world, including Andrew Vanderburg, a visiting assistant professor at MIT, and co-authors from multiple other institutions including the Harvard and Smithsonian Center for Astrophysics, the University of South Queensland, the University of Texas at Austin, and Lund University.

A “one-of-a-kind” system

As their name implies, mini-Neptunes are planets that are less massive than Neptune. They are considered to be gas dwarfs, which are made mostly of gas, with an inner, rocky core. Mini-Neptunes are the most commonly found planet in the Milky Way, though, interestingly, no such world exists in our own solar system. Astronomers have observed many planets circling a wide variety of stars in a range of planetary systems. Mini-Neptunes, then, are generally considered to be garden-variety planets.

But in 2020, Chelsea X. Huang, then a Torres Postdoctoral fellow at MIT (now on the faculty at University of South Queensland), discovered a mini-Neptune in a rare and puzzling circumstance: The planet appeared to be circling its star with an unlikely companion — a hot Jupiter.

The astronomers made their discovery using NASA’s Transiting Exoplanet Survey Satellite (TESS). They analyzed TESS’ measurements of TOI-1130, a star located 190 light years from Earth, and detected signs of a mini-Neptune and a hot Jupiter, orbiting the star every four and eight days respectively.

“This was a one-of-a-kind system,” says Huang. “Hot Jupiters are ‘lonely,’ meaning they don’t have companion planets inside their orbits. They are so massive, and their gravity is so strong, that whatever is inside their orbit just gets scattered away. But somehow, with this hot Jupiter, an inner companion has survived. And that raises questions about how such a system could form.”

A spot-on snapshot

The 2020 discovery of TOI-1130 and its odd planetary pair inspired Huang, Vanderburg, and their colleagues to take a closer look at the planets, and specifically, their atmospheres, with JWST. In its new study, the team reports its analysis of TOI-1130b — the inner-orbiting mini-Neptune.

Catching the planet at just the right time was their first challenge. Most planets circle their star with a regular, predictable period, like the tick of a clock. But the mini-Neptune and the hot Jupiter were found to be in “mean motion resonance,” meaning that each can affect the other’s motion, pulling and tugging, and slightly varying the time each takes to orbit their star. This made it tricky to predict when JWST could get a clear view.

The team, led by Judith Korth of Lund University, assembled as many past observations of the system as they could, and developed a model to predict when each planet would pass by the star at an angle that JWST could observe.

“It was a challenging prediction, and we had to be spot-on,” Barat says.

In the end, the team was able to catch a direct and detailed snapshot of both planets.

“The beauty of JWST is that it does not observe just in one color, but at different colors, or wavelengths,” Barat explains. “And the specific wavelengths that a planet absorbs can tell you a lot about the composition of its atmosphere.”

From JWST’s measurements, the team found that the planet absorbed wavelengths specifically for water, carbon dioxide, sulfur dioxide, and to a lesser degree, methane. These molecules are heavier than hydrogen and helium, which constitute lighter atmospheres. Astronomers had assumed that, if mini-Neptunes formed very close to their star, they should have light atmospheres.

But the team’s new results counter that assumption and offer a new way that mini-Neptunes could form. Since heavier molecules were found in the atmosphere of TOI-1130b, which resides very close to its star, the scientists say the only possible explanation for its composition is that the planet formed much farther out than its current location.

The planet likely accumulated its heavy atmosphere of water and other volatiles such as carbon dioxide and sulfur dioxide in the icy region beyond the star’s frost line. In this much colder environment, water condenses onto bits of dust to form icy pebbles, which an infant planet can draw into its atmosphere. The water evaporates as it slowly migrates in closer to its star.

Barat says the team’s detection of heavy molecules in the atmosphere of TOI-1130b confirms that the planet — and likely its hot Jupiter companion — formed in the outskirts of the system. Through gradual migration, the two planets would be able to stay close together and keep their atmospheres intact.

“This system represents one of the rarest architectures that astronomers have ever found,” Barat says. “The observations of TOI-1130b provide the first hint that such mini-Neptunes that form beyond the water/ice line are indeed present in nature.”

This work was supported, in part, by NASA.

The tech revolution that wasn’t

Peter Dizikes | MIT News — Tue, 05 May 2026 00:00:00 -0400

In 1960, engineers at India’s Tata Institute of Fundamental Research (TIFR) built what they called an “Automatic Calculator,” the country’s first working computer. It had the same type of ferrite-core memory as IBM’s world-leading machines, and at a glance, appeared to herald a new age of tech advances in India.

Constructed with a fraction of the resources Western computer engineers had, the TIFRAC, as they called it, was a remarkable feat.

“The people working on it had never really seen an actual functioning computer,” says Dwai Banerjee, an associate professor of science, technology, and society, and the author of a new book about computing in India. “You had this ambitious group of engineers building a state-of-the-art machine with very, very, limited resources. The fact they could build this is staggering.”

However, the TIFRAC was never even replicated, let alone produced at scale. The visionaries behind it wanted to turn India into an independent computing nation: a place that would produce its own equipment and become an industry power. Instead, the TIFRAC became a technological cul-de-sac, and India’s tech industry took on a very different shape. Instead of exporting equipment, it exports talent, sending skilled engineers and executives around the globe.

Now Banerjee explores those issues in the book, “Computing in the Age of Decolonization: India’s Lost Technological Revolution,” published by Princeton University Press. In it, he examines the country’s pursuit of technological self-sufficiency, and the global forces that prevailed against this vision. As a result, the country is “the world’s leading provider of inexpensive outsourcing and offshoring services, yet enjoys minimal benefits from more profitable advances in research, manufacturing, and development,” Banerjee writes.

“This book is about understanding how the current landscape of technological power came to be and the unequal way in which power is distributed across the world when it comes to anything to do with computing,” Banerjee says. “Basically, the historical conditions of the mid-20th century period are essential to understanding why the world of computing looks the way it does today.”

Computing and the geopolitics of knowledge

When India became a sovereign nation in 1947, many of its leaders believed “rapid technology-driven industrialization was the only way out of centuries of colonial underdevelopment,” as Banerjee writes. Some leapt into action, such as the remarkable nuclear physicist Homi J. Bhabha, who helped establish the TIFR.

Initially, Indian leaders hoped to gain cooperation for the U.S. and international organizations in making technological advances, but quickly ran into Cold War politics. Computing was heavily bound up with defense matters; India was not always fully aligned with U.S. political interests, so the flow of knowledge from the U.S. to India was distinctly limited.

“This is very much an external constraint story,” Banerjee says. “You need blueprints and not just working papers, and that’s what was guarded by the U.S. for a very long time.”

Still, the TIFR research team toiled away as its computing projects until the TIFRAC was up and running — making national headlines.

“The achievement it represents is mind-boggling,” Banerjee emphasizes. “A computer in the U.S. would have cost more to run than this entire institute in India.”

As Banerjee details in the book, the TIFRAC machine was built to grow. Its engineers matched the speed of IBM machines and planned to import larger ferrite-core memory stacks as their workload expanded. But when IBM released the FORTRAN programming language in 1957, it required four times the memory the TIFRAC machine was equipped with. India’s 1958 foreign exchange crisis then shaped the machine’s fate: The World Bank convened a U.S.-led creditor consortium that conditioned rescue loans on the opening of Indian markets to Western capital. Importing larger memory stacks became unaffordable, rendering the TIFRAC obsolete almost as soon as it was completed.

“It’s a geopolitics-of-knowledge question, not that they made a mistake,” Banerjee says of the Indian engineers. “They didn’t know IBM was about to reshape software.”

Exit IBM, enter services

Though IBM’s jump forward after the release of Fortran left the TIFRAC project stalled out, Indian advocates for computer manufacturing did not give up their dream. For one thing, they looked around for partnerships and other ways of moving their domestic tech industry forward. And then in 1978, India, uniquely, banned IBM from the country, on account of its business practices.

That might have set the stage for India’s computer manufacturing industry to flourish. But at the same moment, countervailing forces took hold, including a widespread turn toward the private sector as an increasing source of activity, rather than public-private enterprises.

“For a moment you have this imagination come to a sort of fruition,” Banerjee observes. “But by the late 1970s and 1980s, there is a new group of people arguing for quick profits through software services, saying that this route feels less painful than setting up manufacturing, R&D, and firms for a decade or more.”

This turn toward private-sector services rather than government-involved manufacturing ultimately became a decisive factor in shaping India’s tech-sector trajectory. Rather than seeking to make machines domestically, the country became part of the global tech-services sector, while many of its engineers migrated to Silicon Valley and other tech hotspots. Global tech firms used their reach to advance the idea that many countries would develop independent industries. This is not the outcome India’s leaders and technologists once envisioned.

“It still surprises me because of the one thing India did that no other country in the world managed to do, and that’s kick out IBM,” Banerjee says. “The fact that this vision fades is part of changing government ambition.”

Beyond the mavericks

In writing the book, Banerjee has multiple goals. One is simply shedding more light on the rich details of India’s initial computing efforts. Another is contesting the idea that India somehow naturally found a role providing services and exporting talent; that is not what many people once hoped.

Still another motif in Banerjee’s work is that the history of computing too often centers on innovators who are cast as mavericks, shrugging off conventions to upend business and society — whereas the large-scale forces of global capital and geopolitics matter greatly in technological development.

“This book suggests we often overplay those stories of individual genius, because you can be a genius with all the right ideas, but if you don’t have all the institutions supporting you, it means nothing,” Banerjee says.

Other scholars have praised “Computing in the Age of Decolonization.” Matthew L. Jones, a professor of history at Princeton University, has stated that Banerjee’s book is a “scrupulous accounting of ultimately failed Indian efforts to secure technological sovereignty in the wake of independence,” which “joins the best recent accounts of computing worldwide and transforms how we think through diverse national trajectories through the Cold War and beyond.”

For his part, Banerjee hopes a wide variety of readers will be interested in the book — and recognize that the specific case of India and computing can tell us a lot about the challenges of new types of economic growth in many places.

“India stands in for a lot of countries in the mid-20th century that had recently gained formal political independence and were thinking of ways to catch up with the rest of the advanced industrialized world,” Banerjee says. “But the power structures tied to technological and scientific advancement did not disappear. They were replaced by newer structures, including foreign policy with very specific ideas about what different countries should be doing with regard to technology. That’s where the story starts.”

How chromatin movement helps control gene expression

Anne Trafton | MIT News — Mon, 04 May 2026 05:00:00 -0400

Gene expression is controlled, in part, by the interactions between genes and regulatory elements located along the genome. Those interactions depend on the ability of chromatin — a mix of DNA and proteins — to move around within a crowded space.

In a new study, MIT researchers have measured chromatin movement at timescales ranging from hundreds of microseconds to hours, allowing them to rigorously quantify those dynamics for the first time.

Their analysis revealed that chromatin can exist in two different categories: In one, chromatin moves in a constrained way that allows it to primarily contact only neighboring regions of the genome; in the other, chromatin moves more freely and contacts regions that are farther away, but only over longer timescales.

The findings offer insight into how gene expression is regulated, as well as how chromatin segments come together for other processes such as DNA repair, the researchers say.

“Because we were able to look at chromatin dynamics for the first time at these very fast timescales, and also for the first time across the full dynamic range, we were able to observe chromatin motion over a range that just wasn’t possible before,” says Anders Sejr Hansen, an associate professor of biological engineering at MIT and the senior author of the new study, which appears today in Nature Structural and Molecular Biology.

The paper’s lead authors are MIT postdoc Matteo Mazzocca, Domenic Narducci PhD ’25, and Simon Grosse-Holz PhD ’23. Jessica Matthias, chief commercial officer of Abberior Instruments, and Tatiana Karpova, manager of the National Cancer Institute Optical Microscopy Core, are also authors of the paper.

Constrained movement

In textbooks, chromatin is often depicted as a static structure within the cell nucleus, but in reality, it is constantly moving. Those movements are necessary for genes to interact with DNA regulatory sequences such as enhancers, which can be as far as 1 million base pairs away. They also ensure that when DNA breaks occur, the two ends of DNA can encounter each other to be repaired.

“Chromatin dynamics are foundational to all processes in the nucleus, and especially processes that involve two things finding each other. That’s important in DNA repair, gene regulation, recombination, or moving a particular gene to the right compartment of the nucleus,” Hansen says.

The movement of any particular location on the genome, or locus, is constrained by the fact that DNA is a polymer. After moving in any direction, a locus will be pulled back by the DNA on either side of it.

“Chromosomes are polymers. They’re held together by many nucleotides of DNA. Being part of DNA is a little bit like running while holding hands with other people. If a hundred people are holding hands and you, in the middle of the chain, try to run in one direction, you’ll get pulled back,” Hansen says.

This type of behavior is known as subdiffusive movement. Previous studies have yielded conflicting reports on how subdiffusive chromatin is, mainly because the studies were not able to track the movement over a long enough period of time to obtain statistically robust measurements. Because the movements are so small, on the order of nanometers, data needs to be obtained over long dynamic ranges — from milliseconds to hours.

In those earlier studies, researchers used imaging techniques that can track the position of a single molecule over time by comparing images frame by frame. These are useful but can only be used over a small dynamic range because of the limitations of conventional microscopy.

To generate more statistically robust data, the MIT team used MINFLUX — a super-resolution light microscopy technique that can track the movement of tiny objects such as proteins over longer periods of time. This technique was recently developed by Stefan Hell of the Max Planck Institute, a Nobel laureate for his work in super resolution microscopy. In this study, the MIT team became the first to apply this technique to chromatin in living cells.

“MINFLUX allowed us to get around the limitations of conventional microscopy, letting us measure chromatin movement faster and for a longer period of time than ever before,” Narducci says. “To our knowledge, it’s the first time this technique has been used this way.”

Using MINFLUX, the researchers were able to study cells over timescales that covered four orders of magnitude — from 200 microseconds to 10 seconds. And by combining MINFLUX with two traditional imaging techniques, they could track chromatin movement over seven orders of magnitude across time, from hundreds of microseconds to several hours.

“Region of influence”

These studies, performed across several different mouse and human cell types, allowed the researchers to identify two distinct classes of chromatin dynamics. In both classes, over short and intermediate timescales (up to 200 seconds), any given locus tends to move only within about 200 nanometers. This suggests that the subdiffusive pull is stronger than had been previously thought.

“One of the main takeaways is that you have this region of influence where a genomic locus has access to other genomic loci, and this is roughly a couple hundred nanometers large,” Grosse-Holz says. “If loci are much closer together than a couple hundred nanometers, they’re effectively in contact all the time. You get a cutoff at a couple hundred nanometers where everything within that region around a given locus can see that locus, and everything outside cannot.”

This constant contact is likely beneficial for DNA repair, as the broken strands remain in close proximity to each other. The findings also suggest that for genes and regulatory elements that are within about 100,000 base pairs, they don’t need any extra help to find each other — they will do so routinely through their normal movement.

“If they are closer than 100,000 bases, and most regulatory elements are, then those elements are going to find their target gene within a few milliseconds or a few minutes,” Mazzocca says. “These are timescales that are completely consistent with transcription.”

In the other class of chromatin dynamics that the researchers identified, chromatin is able to move over a wider range, but only at longer timescales (a few minutes to hours). This class of chromatin appeared in some types of cells but not others, for reasons that are not yet understood.

“It would be reasonable to assume that the behavior would be more or less the same in all cell types, but that’s not at all what we found,” Hansen says. “It’s very different in different cell types, with no obvious way of categorizing things.”

He adds that the strength of the subdiffusive pull that the researchers found in this study can’t be explained with existing models that have been developed to study chromatin dynamics — the Rouse model and the fractal globule model. This suggests that the models may need to incorporate factors that were previously left out, such as the interactions between chromatin and the crowded nucleoplasm it sits within.

“These findings are significant for two key reasons,” says Luca Giorgetti, a group leader at the Friedrich Miescher Institute for Biomedical Research in Switzerland, who was not involved in the study. “First, they rigorously confirm longstanding but anecdotal observations that chromatin motion is strongly subdiffusive. Second, they demonstrate that this behavior is consistent across multiple cell types and persists across all measured timescales.”

The research was funded, in part, by the National Institutes of Health, a National Science Foundation CAREER Award, a Pew-Stewart Scholar for Cancer Research Award, and the Bridge Project, a partnership between the Koch Institute for Integrative Cancer Research at MIT and the Dana-Farber/Harvard Cancer Center.