Now we’ll see whether it works, specialists on technology law say.

Meta decided on Aug. 26 to settle a lawsuit brought by dozens of state attorneys general for nearly $18 billion over claims that it endangered the physical and mental health of children and teens with its addictive social media platforms, including Facebook and Instagram.

In addition to the payout, Meta agreed to modify a number of design features that critics and researchers say encourage excessive use as well as those that psychologists say lead users to making unhealthy negative social comparisons between themselves and peers or influencers.

The recent Meta case was being closely watched by legal experts and is considered a bellwether. But legal scholars note that it is just one of thousands still pending against social media companies.

In this edited discussion, the Gazette asked Leah Plunkett ’01, J.D. ’06, Meyer Research Lecturer on Law at Harvard Law School and a faculty associate at the Berkman Klein Center (BKC), and Christian Sandvig, a BKC fellow who studies the ethics of information technology and is part of the Center’s new AI & Youth Safety Initiative, about their take on the agreement.

---

Gazette: How big of a deal is this settlement?

Leah Plunkett: This is a huge deal. This is a paradigm-shifting deal when it comes to social media and the way it interacts with kid and teen users, as well as the family, schools, and communities around them. It is also paradigm-shifting in terms of what those kids, teens, families, schools, and communities can expect in their interactions.

I think it sends a huge warning shot across the bow for any AI or other emerging tech company. If they choose to take the same push-the-boundaries, see-if-you-can-get-away-with-it approach that the social media companies have by and large taken, it’s going to be trillions with a T instead of billions with a B when the tab is called for them because of the size and scale they’re operating in.

The other thing about this is we see the states decisively establish themselves as the trusted leaders for the public interest when it comes to kids, teens, schools, families, communities, and digital life. The federal government has not succeeded in providing the type of comprehensive statutory or regulatory reform, or even the sustained and creative types of enforcement actions that are needed in such a complicated tech landscape.

Christian Sandvig: It’s great news and it’s terrible news at the same time. I was happy about it, but there are some serious problems with it.

I would say there’s a consensus that the platforms are causing harm, and it’s always nice to see a settlement or some sort of judicial process where people who have been wronged are made good. So, generally speaking, it seems like steps in the right direction.

I was pleased that there are interface changes that have to be done by the platforms. Some are straightforward and obvious, things that, in the trial, the whistleblower was identifying as necessary. That’s great that they’re doing those things as opposed to just the money.

The period of lockout — 12 a.m. to 6 a.m. — is really well-supported by evidence. Researchers have found that sleep disruption is a big consequence of using these apps. So, if you had children who are bound to the account, and there’s a way to enforce that it’s really a child’s account, then that seems like a positive move.

But the negative side of it is — just to take one example — Facebook introduced the “like” button in 2009. Academic researchers were writing about how this could be harmful to adolescents as early as 2013. I wrote a paper about it in 2013, and it’s 2026.

And so, if this is our template for regulating technology, it failed. We didn’t do anything about it for such a huge number of years.

The other negative is it only affects Meta; the other apps don’t have to do anything. Even if it just affects Meta and it’s good, we still have some questions about whether the specific features that have been proposed are going to work.

For an example there’s an exception for direct messaging. But we know that direct messaging is a source of all kinds of horrible things that youth and teens have to experience on social media, like bullying and sexual imagery that is unwanted.

By settling, is Meta essentially conceding that some of these features are potentially harmful to children and teens?

Plunkett: What I see Meta doing here is while preserving, as a legal matter, their position that they’re not admitting wrongdoing, as a practical matter, and certainly as an optics matter, for the public this is Meta behaving as if they know that major features and operations of their platforms have been risky and harmful to kids and teens, and by extension, their families, schools, and communities.

Or at minimum, they are unable to establish convincingly, both in court and in the court of public opinion, that these major design and operations features of their platforms were safe or neutral.

Are any of the feature modifications Meta has agreed to make, such as limiting the amount of time and time of day younger users can be on the platforms, likely to make a difference?

Plunkett: I’ve been reading through the 122-page settlement document. What I am seeing so far is we have age assurance, which is a necessary, but not in and of itself sufficient, condition to provide protections. Age assurance is a bit like coming up with a way to ID at the front door of a restaurant and see if somebody gets the drink bracelet or not. It does not act as a safety protection; it’s identification.

I’m seeing what the settlement agreement is calling “time management,” meaning overall duration of time allowed per day, night access mode 10 p.m. to 7 a.m. based on the device’s local time zone.

Push notifications will be disabled, which is a big deal. “Meta must implement productive pauses and notices to reduce or prevent excessive mindless or unintended teen usage.” I think that may almost matter more than the hours, because core to Meta’s business model is directing and capturing focus and attention. I do think that matters.

The settlement agreement also has Meta agreeing to “identification of suspected secondary accounts.” It looks to me that this is designed to help try to cut back on teens having multiple accounts. But, whether it’s Meta or any other social media platform, a teen with a real commitment to do an end run could figure it out. There are any number of ways that teens could still try to get around this.

A lot of people are talking about this settlement as being a Big Tobacco moment. That is true in a descriptive way. My understanding from the state AGs’ press releases is that this is the largest state consumer protection settlement in history outside of the big tobacco settlements of the ’90s just in terms of money.

What still isn’t being done that you think would make a difference?

Plunkett: I think it is really significant that Meta is taking these steps around what they’re calling the social comparison tools. These types of tools have documented impacts on young people’s health and sense of self and emerging identity.

The reason it’s so important is that these kinds of social comparison filters are in a somewhat odd or even sensitive place from a legal perspective in terms of whether they are speech or activities by the platform or by companies authorized to be on the platform that might enjoy some type of free speech or even Section 230 protection such that it would be difficult, if not impossible, for a federal or state legislature or a federal or state regulator to ban them or limit them.

So, to have the platform itself say, “we’re dialing these back” is really the best way for this to be addressed, because it gets us out of some of these complicated and still very much emerging and unsettled questions.

There’s a lot in the injunctive relief part of the settlement that is designed to limit or prohibit teens from being contacted by inappropriate other users or seeing inappropriate content. There’s better parental safety and supervision; there’s better content reporting. If that all happens, that’s really big.

What I would like to see Meta and the other platforms do as well is put ethical, practical limitations in place on the marketplaces that exist for child and teen influencer content that draw adult audiences, particularly adult audiences who have — to use the language of the settlement — inappropriate intentions or, in some instances, illegal or criminal intentions.

I do think that there still are a lot of marketplaces on Meta and other social media platforms where we are seeing parents or other adults monetize and, in too many instances, exploit the private experiences of or performances by their kids or kids in their lives for the viewing pleasure, advertising dollars, brand sponsorships, of other adults and institutions.

This settlement does not look to me like it is going to meaningfully disrupt and regulate those entertainment marketplaces. That remains a pretty big area of my concern. But that was not the focus of the AGs’ litigation.

Sandvig: The issue that researchers have pointed out is it’s fundamental to the technology, and so it’s not clear how to solve them with features. Researchers have complained that many of the safety features already in all social media platforms are ineffective. They are usually settings, and they’re hard to find.

In this settlement, there are instances of changing things that used to be opt-in to things that affect everybody or all children on the platform. So, it’s clear that someone was listening to that criticism.

But then there are still other features that were implemented as options, like the algorithmic feed versus chronological feed. That’s an option that no one would use, I think.

This isn’t probably something that we should solve with repeated lawsuits, but rather some structure of regulation. Political commentators say this is a bipartisan issue where we have consensus, which is very rare. That there’s a lot of people in both parties who are excited about doing something about this. I guess the downside of the settlement is perhaps it makes that harder because people might think the problem has been solved.

As Harvard’s Class of 2030 arrived in Harvard Yard for move-in day, students shared what they’re looking forward to, from meeting new people to starting classes to exploring campus, while their parents talked about their own mixed emotions. They shared feeling both pride and sadness that comes with the day, and the start of a new chapter — for everyone.

Harvard Yard awoke from its summer slumber on Wednesday as members of the Class of 2030 were greeted by campus regulars eager to welcome the first-year students to their new home.

Among the fresh faces was Nathaniel Shugarman, who arrived from the Florida Keys.

“This is incredible, to be here,” said Shugarman. “I’m definitely nervous about the classes and adjusting from high school, but I’m excited for all the people. It’s incredible. It feels so historic and electric.”

His mother, Megan Shugarman, described her emotions.

“I’m feeling super blessed and grateful. It’s the end of a childhood era, so that part makes me sad. But I’m so excited, and everything’s been so friendly and open.”

Also moving a plane ride away from her family was Bethany King, a water polo player from Santa Barbara, California.

“I’m pumped. I’ve been greeted with lots of vibes, high energy. It’s getting me excited for my four years,” she said, grinning alongside her parents, Sheryl and Brian. “If this is my first day at Harvard, shoot, what’s going to be in the next four years?”

King said already, she’s making tons of friends.

“It’s just like Instagram requests,” she said. “Follow, follow, follow, follow.”

Moving closer to home was Ilaria Luna, whose father, Franceso, and sister Esther patiently waited with her belongings outside Wigglesworth Hall. Harvard is closer to their hometown of Washington, D.C., than Shanghai, where Luna just finished a gap year.

“We were empty-nesters last year,” Francesco said. “So we’ve been through this.”

Likewise Thomas Grieser’s parents, who live in Newton, Massachusetts. Thomas spent the past five years attending Eton College in the United Kingdom.

“I’m excited to have him closer to home,” his mother, Kathleen, said as the pair took a break from moving boxes into Straus Hall. “And my husband and I both went to Harvard, so it really feels like a homecoming.”

Outside Matthews Hall, Leah Wurzburger guarded her daughter Hannah’s pile of belongings while helpers lugged her first load up several flights. Watching the third of her three kids move into College was bittersweet, she said.

“She’s my baby,” she said. “It’s very emotional, but I’m very excited for her. I’m very excited for all the opportunities.”

And while this wasn’t her first college move-in rodeo, it was her first with a daughter, Wurzburger said. To sum up her feelings becoming an empty-nester, she quoted Stephen Sondheim lyrics.

“I’m excited and scared,” she said. “Hannah’s a little bit nervous, mainly excited, and I’m all those things plus very sad.”

President Alan Garber was out in the Yard, greeting students and parents.

“I get a huge burst of energy meeting each incoming student, and from all I’ve heard and what I’ve seen today, this will be an incredible group of Harvard students,” Garber said.

Humans didn’t evolve to shop for food that has been collected, packaged with nutritional information, and arranged by category in brightly lit supermarkets. In fact, we didn’t evolve to choose our food at all. 

In his new book, “Fed Up: What Evolution Reveals about Food, Diet, Health, and Eating Well,” Daniel Lieberman, Edwin M. Lerner II Professor of Biological Sciences in the Department of Human Evolutionary Biology, explores how our ancestors evolved to eat almost anything they could hunt, gather, or grow in whatever habitat they happened to find themselves. Which makes the modern American diet’s embarrassment of riches a real conundrum. Should we eat low-carb or low-fat, vegan or strictly animal products?

In “Fed Up,” Lieberman argues that most diets cling too firmly to one “big idea” and that they can, and should be, much more flexible. Lieberman helped us develop this quiz based on the book to help us make sense of it all.

---

Researchers with the long-running Harvard-Cornell Archaeological Exploration of Sardis have unearthed a larger-than-life statue, estimated to be nearly 2,500 years old. 

The discovery is remarkable for several reasons, said Nicholas Cahill, the field director of the AES team and a professor of Greek and Roman art and archaeology at the University of Wisconsin-Madison. 

“It is very rare to find such monumental sculptures of this period anywhere in the Mediterranean, and this exceptionally well-preserved, unique monument casts important new light on the major arts of Lydia, and on the Lydians’ ties to their own past,” he said.

The excavation is sponsored by the Ministry of Culture and Tourism of the Republic of Türkiye and by Harvard Art Museums. The Turkish site was named a UNESCO World Heritage Site last year, cementing its position as a critical site for understanding the ancient world.

Researchers made the discovery in late May while examining a major Roman colonnaded avenue. The statue had been broken in two main pieces and reused, face down, as paving stones in the ancient thoroughfare. 

“You sort of see these whole worlds opening up in front of you,” Cahill said of the moment he realized the team had found something extraordinary. 

A team of conservators carefully excavated around the statue to gradually reveal its full shape and size in the ground.

“Then the guesswork started,” said Susanne Ebbinghaus, the George M.A. Hanfmann Curator of Ancient Art at the Harvard Art Museums, who was in Sardis for part of the excavation. “‘Is this archaic? Maybe not, it looks a little later. Is it a guy? Is it a girl?’  When they started cleaning the soil around the head, we saw he had sideburns, so that question was settled.” 

The people of the Lydian empire, which controlled what is now western Türkiye from about the middle of the 7th century to the middle of the 6th century B.C., left little written record of their civilization. Much of what was known came from what Greeks and Romans wrote about it — and from their point of view, the Lydians were eastern barbarians. But archaeological evidence has revealed that Sardis, the capital of the Lydian empire, was a center for luxury and culture, particularly perfumes and fine clothing.

James McCurley ’27, a Classics and Physics concentrator who worked on the dig over the summer, said the excavation of the mammoth statue proved the highlight of his experience. “After all the anticipation, I was worried the statue might be underwhelming, that we’d dig it up and find its front weathered beyond recognition. Discovering that the statue was not only well-preserved, but almost intact and 2,500 years old was a wonderful surprise.”

Conservators lifted the pieces from the ground on July 13 and transported them to an AES conservation lab to begin a thorough, and still ongoing, analysis. 

Many details have been revealed since that early, excited guesswork. 

At well over 7 feet tall, the figure would have towered over onlookers. He wears a tunic of thinner material (called a chiton in Greek), a diagonally wrapped mantle, and a distinctive, perhaps local garment with fine horizontal folds. The design has few parallels in ancient sculpture, said Harvard Art Museums’ Bahadir Yildirim, who is the Sardis expedition administrative director and who participated in the cleaning and analysis.

“The surprising thing about this piece is that it has some features that haven’t been seen on sculpture like that … the drapery underneath, the distinctive horizontal patterning,” he said.

The statue’s bolt-upright stance, long coiffed hair, and manner of dress resemble archaic Greek sculptures from the 6th century B.C. But the prominent veins and other anatomical details betray a much later date, likely in the 470s to 450s B.C., the early Classical period in Greece. During this era, Sardis was no longer an independent capital but had been conquered by the Achaemenid Persians and had become a major provincial center. 

But the figure shows no Persian influence. The statue, therefore, comes across at first as a very traditional sculpture, adhering to age-old hairstyles and dress codes, while also hinting at familiarity with the current styles in Greece where such a sculpture might have been depicted as a nude youth. But the Lydians and other Anatolian peoples did not exercise naked and did not depict their young men without clothes.

The references to multiple cultural traditions and time periods reflect an artist with a complex understanding of his or her place in the world, Cahill said. 

“This sculptor knows about contemporary styles in Mediterranean and Anatolian sculpture, the interest in hands and the structure of the face and so forth, and yet [the statue is] wearing clothes from 100 years earlier,” Cahill said. “I think they’re very clearly looking back to their own past.” 

The original placement and purpose of the monumental statue is unknown. But one possibility is that it stood at the sanctuary of Cybele, the native mother goddess, which has never been found but has been described in literary sources as one of the most important sanctuaries in Sardis. It was common for wealthy citizens or political leaders to ensconce their likeness in the presence of a deity so they could effectively worship in perpetuity, explained Ebbinghaus. 

“You put yourself in a prominent place where your fellow citizens or subjects can see you, but also you are indeed worshiping the god — and if you’re a ruler, drawing legitimization from the relationship to that deity. So it might well have been an important person being represented,” she said. 

Most ancient marble statues were painted, and this one was likely no exception. The Sardis conservation team is now working to clean the statue while preserving any remaining flecks of paint for color analysis. 

When conservation is complete, the fragments will be joined and the statue will be displayed in a Turkish archaeological museum, along with other finds from Sardis. The first Harvard excavation of Sardis, a joint project with Cornell University, began in 1958 and was led by George M.A. Hanfmann (1911-1986), a professor of fine arts at Harvard and curator of ancient art at the Fogg Art Museum.

Hungarian Prime Minister Viktor Orbán’s 16-year rule ended in the spring, when voters replaced him with Péter Magyar, who has pledged to undo Orbán-led initiatives that in the eyes of critics isolated and politicized science.

Albert-László Barabási, a Harvard Medical School lecturer on medicine, part-time, at Brigham and Women’s Hospital and a distinguished university professor at Northeastern University, is a graduate of Eötvös Loránd University in Budapest and a visiting professor at the Central European University in Vienna. He recently penned an op-ed in the journal Science laying out a blueprint for how Magyar might proceed.

We spoke to Barabási about his proposals for Hungary and how politics can undermine research. The interview has been edited for clarity and length.

---

Gazette: What happened to Hungarian science during the Orbán administration?

Albert-László Barabási: Hungary has a particular pride in science and in how many intellectuals the country produces, particularly how many Nobel Prize winners, which is close to a dozen. That includes two in 2023, one in physics and one in medicine, and last year’s literature winner.

The Orbán regime politicized science. Research institutions and national labs had traditionally been under the direction of the Hungarian Academy of Sciences. About seven years ago, the government removed these institutions from the academy, saying they wanted to reorganize them to be more competitive. Virtually all scientists, including all members of the academy, were against the change, but it did not matter.

To their credit, the government attracted a well-respected academic back from Singapore to run the new institutions, and increased funding. They increased both salaries and the resources going to research projects, but they never managed to address the psychological break that resulted from stripping away academic control.

It did not end there. The Orbán government started a system that, on one hand, you could view as a positive. But on the other, the execution was politicized. They wanted to bring the Hungarian universities up to an international level and concluded, rightly, that the universities needed new funding and independence. So, they created foundations that effectively made them into organizations that are independent of the government and then donated state properties — shares of major companies, land — into those foundations to secure the resources to expand and function.

In principle, this makes lots of sense. In practice, the boards were stacked with political appointees, so they were not independent. As a result, the EU protested and locked out these universities from the Erasmus program, which supports education, as well as from the European funding system. The few universities that did not undergo this transformation remained within the EU system, but, deprived of new funds, became deeply impoverished.

What was the impact on the quality of the country’s science?

Science is a collaborative enterprise, so no matter where you are in your career, undergraduate, Ph.D. student, postdoc, even faculty, you need to be able to collaborate and learn from international colleagues. The EU has created a framework for that, and Hungary had been successful at taking advantage of it. That framework was artificially cut — students couldn’t travel, faculty couldn’t collaborate — and the consequences will probably be seen decades from now, in lost collaborations, diminished impact and productivity.

You outlined several steps that Hungary’s new leaders should take. First is re-establishing trust. What do you mean by that?

The Orbán regime lost trust by packing the university boards in a way that wasn’t transparent, and by removing the control of the research institutions from the academy. There is a need for a reset — and it is already happening. While the outcomes are uncertain, the direction is good. In my mind, the question is, will the universities retain independence and get the funds they need to thrive? That we can probably only assess later.

Another recommendation is to make reforms durable.

The independence of the research network is technically already in place and guaranteed by law. The question is whether the new government agrees with that setup or whether they’ll seek an alternative. It doesn’t really matter in my mind whether it’s controlled by the academy or by the government or whether it’s set up as an independent body. What matters is the independence of the researchers and the decision makers who fund the research.

Have the decision-making structures in the U.S. been damaged in a similar way, in your view?

We have seen evidence of the decisions of the scientific body being overturned by political appointees, both in the U.S. and in Hungary before the elections. Who gets funded or who does not get funded is no longer a decision left to the research community alone. When I talk about independence in the U.S., I mean restoring a system that for decades had no political interference. All that matters is if my research is of sufficient quality, not whether it is politically acceptable.

Hungary seems to be on the road to de-politicizing its science. How long is this road?

You could say six months: Put the laws in place, secure the necessary funding, and let it go. But the government has many other issues to deal with. There are issues with the healthcare system, the educational system, and outdated transportation. So how do you assign priorities? I was in Budapest in May at the National Academy meeting, and the new prime minister came and addressed the membership. He said that he intends to increase funding to the European norm, which requires allotting another 1 percent of GDP to innovation and science. If he lives up to those words, Hungary could open another golden era in science.

Why is there sex — or, more precisely, sexual reproduction? If you’re an evolutionary biologist, the answer isn’t obvious.

Sexual reproduction is biologically costly, requiring not only competition for mates, but also the creation of entire populations in which only half the individuals can give birth. That reality has led scientists to wonder why it arose in the first place. After all, in asexual reproduction — as occurs when bacteria divide — not only can you skip the wooing and its costs, but every individual is a potential mother.

New research from a group of Harvard biologists dives into the benefits that come with sexual reproduction, particularly in changing environments.

The work was conducted in the lab of Fisher Professor of the Natural Sciences Michael Desai and led by Ph.D. student Shreyas Pai and Parris Humphrey, then a postdoctoral fellow. The team used a strain of yeast as a model organism. The yeast, commonly used in bread baking and beer brewing, has the scientific advantage of being able to reproduce either asexually or sexually.

Another advantage is that yeast produce new generations in only 90 minutes, taking the guesswork out of experiments on evolution, which is generally understood to promote slow change over long stretches of time. With a new generation every hour and half, Desai’s team was able to observe changes after 960 generations in just four months, something that would take hundreds to thousands of years in large mammals.

The research, published in the journal Science in April and supported by the National Science Foundation, the National Institutes of Health, and the National Institute of General Medical Sciences, began with a dozen identical populations of yeast growing in ideal conditions. All initially reproduced asexually, but every 100 generations, designated lines were induced to reproduce sexually for a single generation.

The scientists first confirmed earlier work showing that in a constant, favorable environment, sexually reproducing populations gain more fitness than asexually reproducing populations. Over those nearly 1,000 generations, the fitness of sexually reproducing populations increased 8 percent over the ancestral population, compared with a 5.7 percent increase for asexual populations.

Next, the researchers tested evolution in different environments — one saltier, one hotter, another with decreased acidity, and one with lower phosphate. In these altered conditions, researchers found that sexually reproducing populations had fitness advantages of between 2 percent and 5.6 percent over asexual populations.

While those numbers may not seem large enough to matter, Desai said that over long periods of time, an advantage of even a fraction of a percentage point can allow a particular population to dominate.

“Selective differences of fractions of a percent in large populations can drive total shifts in population composition,” said Desai, the paper’s senior author. “A several percent effect in a microbe is large. It might take 50 to 100 generations to take over.”

Desai said that the work also probed the mechanism by which sexual reproduction is advantageous in a shifting environment. Some genes are “pleiotropic,” meaning that they affect the expression of several traits, some of which can be adaptive, some harmful. Harmful traits that affect an individual’s ability to survive or reproduce are reduced or removed from the population naturally. But traits that are neutral or only mildly harmful can survive as what scientists call “hitchhiking load,” particularly if other traits of the same gene provide an adaptive advantage.

When the environment changes for the worse, as it did for the experimental yeast, some of those previously neutral or mildly disadvantageous mutations may threaten survival. That is where sexual reproduction gets its edge.

Asexual populations have no way to eliminate the threat because their offspring are clones and genetically identical. Sexually reproducing populations, on the other hand, produce offspring with a mix of their parents’ genes, some of which may not carry the harmful “hitchhikers.” Over time, that reduces or eliminates the harmful effects of those genes.

“We showed that, without sex, it’s not just that the populations are accumulating deleterious baggage, they’re accumulating baggage that maybe is not so bad in the environment they’re evolving in, but would hurt you elsewhere,” Desai said. “In contrast, when you have sex, you purge that hitchhiking pleiotropic cost and prevent yourself from specializing quite as much just to the environment that you’re in. That can be a big advantage when the environment shifts or fluctuates.”

While Desai and colleagues worked with microbes, he said that the effects extend into more complex organisms where asexual reproduction, while rarer, is still represented. In the plant kingdom, for example, cuttings can grow into fully mature plants that are genetically identical to the parent. Among animals, certain species of lizard can undergo parthenogenesis, in which their eggs are self-fertilized, producing offspring that are clones of their parents.

“We believe that this general mechanism by which sex speeds adaptation and purges hitchhiking deleterious load should be a general advantage of sex across all sorts of organisms,” Desai said. “I think it’s great that we can learn so much about how evolution acts and how the ways in which we reproduce, copy, and re-sort our genomes makes sense all the way across these enormous differences in organisms’ complexity and scale.” 

On their first day of preparation for “Pop Culture Jeopardy!,” contestants Jack Goldfisher and Henry Johnson realized they had a problem. Neither of them had watched much television.

The 2018 College graduates had met in the course “Men to Boys: Masculinity in Postwar Hollywood Film,” where many of their classmates had encyclopedic knowledge of both forms. Now the pair were living in Los Angeles, and Goldfisher was sure Johnson — who had co-written a kids’ show for Apple TV — would have the category handled. It turned out they both knew much more about film.

“We thought that each other knew a lot of information because we know a lot of the same information,” Goldfisher said, “which is probably not the way to optimize the team.”

Goldfisher and Johnson hadn’t planned to compete as a duo. When they were selected for the show, they didn’t even know each other very well. Their mutual friend Jack Stovitz, who took the same course under Professor of the Practice in Media and Activism Michael Bronski, had recruited them to a three-person team before the first season of the “Jeopardy!” spinoff. When the show cut team sizes by one for its second season, Stovitz, who had just sold a script to Netflix, bowed out for convenience and left the acquaintances to fend for themselves.

On a large screen in Goldfisher’s apartment, the “AMC B-Listers,” as Goldfisher and Johnson would be known, began to prepare. They took online quizzes on their weak subjects, like anime and country music, while reviewing the results of decades of major award shows.

The goal was not mastery but modest familiarity. They skipped detailed synopses on shows like “The Wire” and practiced identifying the two main characters, the setting, and the theme song. They created several dozen flashcards, many of which just included names of bands and all of their members. Season 2 of “Pop Culture Jeopardy!” was set to stream on Netflix, so they spent extra time reviewing content on that platform, just in case.

On the first day of filming, the two showed up early in the morning at Sony’s Culver City lot not knowing when they might compete in the five episodes recorded that day. When they weren’t called for game one, they waited in the green room with teams who all seemed to know the answer to every question asked by host and 2004 Harvard College graduate Colin Jost.

“Well, everyone here seems to know everything,” Goldfisher said, “so we were like, whatever happens, happens.” Their goal was to win one game before losing — “because you inevitably lose,” he added, “which is the worst part about ‘Jeopardy!’”

They felt lucky that their jitters had subsided by the time they were selected for the day’s final taping. When their educated guesses led to a near-sweep of a category about the pop culture podcast “The Breakfast Club,” they realized that aggression would play a big part in their success.

Not knowing everything about television was fine; in “Final Jeopardy!,” each knew half the background knowledge required to identify which ’80s song became a top-selling digital hit decades after its release, thanks to its use in two series finales. (“Don’t Stop Believin’” from “The Sopranos” and “Glee.”)

After the relatively narrow win in the first game, Goldfisher and Johnson dominated the second and third, clinching a spot in the season’s final tournament. They attributed part of their success simply to enjoying the game and not burdening themselves with expectations. “I think Henry and I both love games so much that we’re able to separate the stakes from the game,” Goldfisher said. “We were having fun. I think that’s a big element of it.”

A large part of the fun was becoming closer friends. On their day off before the playoffs, they had to keep their success a secret. “It’s not like a sporting event where the game is airing that night and everyone can see how you do,” Johnson said. “The fact that we had to go through this really condensed boot-camp period where we’re obsessively excited about this thing that no one else knows about is a really fun bonding experience.”

In the semifinals, the pair found themselves in a distant second place after the first round. “I think, looking back, we probably could have prepared better,” Johnson told Jost between rounds. “I was really convinced there’d be a category about Borscht Belt comedians.”

They were kept in the game partly through their dominance of the “Real Housewives Wit and Wisdom” category, about a show they didn’t watch. Entering Final Jeopardy!, they were in second place. The last prompt: “The last to garner nominations in all four major acting categories at the Oscars were both directed by him.” For 30 seconds, the other groups struggled, but Johnson knew the answer cold. Naming “David O. Russell” earned them a spot in the finals. As the camera flashed to them, Goldfisher gave Johnson a kiss on the cheek.

The championship didn’t start so well. After part one of the two-episode final, they fell nearly 30,000 points behind the dominant “One Baddie After Another” team. Before game two, they agreed they’d have to dial up the aggression. Every guess deserved a buzz. Botching a couple of early questions only encouraged them; they’d have to risk making mistakes if they wanted a chance to win.

Tides started to turn when they doubled up their earnings on a Daily Double! question about the actress Awkwafina. Then, right before their second Daily Double, they nailed a question about Balatro, a game their friends had frequently played until 3 a.m. “It felt like an enchanted moment,” Goldfisher said — like a scene from “Slumdog Millionaire.”

Still nearly 20,000 points behind their competitors, they risked everything to double their 21,200 total on the category “Headlining the Festival.”

In 2022 J. Cole had a good time headlining this Tennessee musical festival whose name is Creole slang for “a real good time.”

Goldfisher, focusing on the rapper J. Cole, thought back to the music festival the rapper’s label put on in North Carolina. But that didn’t seem correct.

“It can’t just be Bonnaroo, right?” Johnson whispered to his partner. It felt too obvious. The major festival takes place in Tennessee but didn’t hold an exceptionally close tie to the rapper.

“If we’d had a semi-reasonable alternative to Bonnaroo, we probably would have guessed it,” Johnson said, “but we came up so empty that it was like, ‘We have to say a word.’”

Bonnaroo was right. The two pushed into the lead as the crowd, and even one opposing team, broke into applause.

“I told you anything could happen!” Jost exclaimed.

The duo couldn’t have been disappointed when the category of Final Jeopardy! popped up: 21st-century movies.

Edith Macefield’s refusal to sell her house to developers, who built around it, was said to inspire the makers of this Oscar-winning film.

The two baseball fans plan to use some of the money to attend a Cubs vs. Mariners game together. Goldfisher has already put some of the money toward completing a low-budget feature he co-wrote and directed this February, and Johnson plans to put money toward the scholarship fund of a high-school classmate who died.

It’s Sunday morning, and everything seems normal in the small town — until a peculiar transformation happens: One by one, the townspeople turn into rhinoceroses.

So begins the first production of the American Repertory Theater’s 2026-2027 season, an adaptation based on Derek Prouse’s English translation of the Eugène Ionesco absurdist play “Rhinoceros,” which premiered in 1959. Directed by the Terrie and Bradley Bloom Artistic Director Diane Paulus ’88, “Rhinoceros” features three Emmy Award-winning actors: Paul Giamatti, Tatiana Maslany, and John Turturro. The show opens Friday and runs through Sept. 27.

“It’s a serious play, but there’s something very childlike about it,” said Giamatti, known for his Golden Globe-winning role in “The Holdovers” and HBO’s SAG-winning “Too Big to Fail.” He plays the pompous, bourgeois character Jean, one of the first residents to transform. “The more bizarre and funnier it is, almost the more serious it becomes in a strange way,” he said.

The premise of the play is slightly ridiculous. Yet it allows the audience to explore what it means to be human in a world that values conformity. That theme attracted Turturro, the award-winning actor, writer, and director, because of how applicable it is today, he said. He plays the lead role of Berenger, who must decide whether to follow the crowd or remain true to his values.

“There are certain essential themes in literature or in the world that just recycle,” he noted. In “Rhinoceros,” Turturro sees how the characters become afraid of their differences — so much so they are willing to accept the flawed logic that eventually leads to “barbaric” behaviors. “[What starts as] a small discrepancy between people becomes this big, giant chasm,” he said.

Tony Award-winner Paulus said that when Ionesco wrote the play, he used metaphor to speak to his experience growing up in Romania and witnessing the Iron Guard, seeing the rise of Nazism in France, and then watching the French intellectual left embrace Joseph Stalin. Paulus believes the story will resonate with modern audiences.

“This play allows us space to look at things that are going to feel like they’re ripped out of the headlines, and yet we’re talking about rhinoceroses taking over a little, small town,” she said. “It allows us to hear the play in a timeless way and also allows us to relate to it very personally.”

When Maslany — known for her roles in the television shows “Orphan Black,” “She-Hulk,” and “Maximum Pleasure Guaranteed” — read the play for the first time, she was floored.

“I read it in one sitting and just kind of felt bowled over by it. It really hit me in this incredibly emotional way,” she said. She plays the “confusing and complicated” character Daisy, an office receptionist and Berenger’s romantic interest. “It was really funny and felt like it couldn’t have been more prescient.”

Giamatti said the play raises fundamental questions about humanity amid the rise of social media and artificial intelligence.

“What makes you a human being, what makes you an individual?” Giamatti said. “[Berenger is] the one guy who’s fighting to be an individual, and he’s a wreck. And what makes him human is the fact that he’s a wreck.”

The cast also includes Harriett D. Foy, Peter Francis James, Kristine Nielsen, Alex Ross, and Clay Singer, with Britt Faulkner, David Mason, Amelia Sipkin,and Scott Stangland as understudies. Everyone has worked together to give the historic play new life.

“The way that [Turturro] leads is [by] facilitating what’s beautiful about every performer that’s in the space,” Maslany said. “He sees something, and he’s like, ‘Oh yes! Let’s invite that more into the show. Let’s, in fact, highlight it and make it a thing.’ It’s so infinitely generous.”

“Rhinoceros” is Paulus’s last directing project at the Loeb Drama Center before the A.R.T. moves to its new home in Allston, the Goel Center for Creativity & Performance.

“It’s your dream as a director to be working with actors like this cast, who bring the combination of craft and attention and rigor with utter playfulness and creativity and spontaneity,” she said. “It’s been an incredible collaboration in discovering what this is [and what] will be unique to our production.”

As for the actual mechanics of transforming into a rhinoceros on stage? Well, the cast says you’ll have to see it for yourself.

“You have to buy a ticket,” said Giamatti, who added he has always loved stories about people “unlacing and letting their animal out.” He’s excited to give it a try himself.

“It’s fun, it’s interesting,” he said. “Just as an acting thing, it’s a challenge.”

For more information and to buy tickets, visit the A.R.T. website.

Finding a calling wasn’t the hard part for Jon Star. That came when he started figuring out the complexities of his chosen profession.

Growing up in Columbus, Georgia, Star worked as a camp counselor and a math tutor, and the jobs sparked something in him. He liked helping kids learn.

His path became clearer years later during a stint as a math teacher at Albuquerque Academy in New Mexico.

Star was tutoring a smart, hardworking student who came from a family where everybody excelled at math, except her. Trying to help this student work through (and sometimes around) roadblocks sparked questions for Star about how math is taught and learned.

 “The student ended up being successful in the course, but I’m not sure how much that had to do with me or her hard work,” recalled Star, now the Carl H. Pforzheimer, Jr. Professor of Teaching and Learning. “That experience made me realize that learning math is a complicated process in which the mind of the student and the actions of the teacher have to come together.”

Star, who has studied how children learn math in middle and high school over the past 20 years, says one of the most effective ways to teach math is by helping students figure out how to solve math problems using multiple strategies, rather than a single method.

In a research paper co-authored with researchers from Vanderbilt University on algebra instruction, Star found that math teaching and learning improve when students begin to develop flexibility in problem solving, a concept that involves the knowledge of several strategies and the ability to choose among them on a particular problem.

“There are many ways to solve a math problem,” Star said. “There is never one way.”

In fact, that observation highlights the problem with the way math was — and sometimes still is — taught. By rote.

“We often think that the best way to teach someone how to solve a math problem is to show them one way to do it, including the steps, and then have them practice that one way to do it,” said Star. “What I’m after is not just, ‘Did you get the right answer?’ or ‘Did you follow the steps?’ but it’s so much broader, richer and deeper than that. … Learning to solve problems in multiple ways helps them understand the math, but also to be able to solve unfamiliar problems when they encounter them.”

Math teaching in the U.S. has improved over the years, said Star, but it has yet to translate into students’ math scores, which lag behind pre-pandemic scores as well as those of many peer nations.

According to the latest Nation’s Report Card released in 2024, the average math score for fourth-graders was two points higher than in 2022, but three points lower than the 2019 pre-pandemic score. The 2023 results from the Trends in International Mathematics and Science Study (TIMSS) showed that U.S. fourth-graders’ math scores were better than the international average but worse than those of Singapore, Japan, England, Sweden, and Germany, and several other countries.

Among the factors that can complicate math teaching are emerging technologies — AI in particular — along with dwindling investment in teacher training and mounting shortages of practitioners, said Star. He worries that math teachers may be choosing to teach in ways that are easier for them but may not be the best for students.

“Teachers are facing a difficult choice,” said Star. “The questions are how do we create incentives in the teaching profession that lead teachers to teach in a more ambitious way, and how can we maintain high standards in teacher training so that teachers are trained to teach the way we know through 30 years of work that teachers should be teaching?”

Math teachers should see themselves as discussion guides, not merely lecturers explaining concepts, said Star.

“There’s certainly going to be times when they’re going to teach, but there are other times when they need to stimulate a discussion and mold it into the outcome that they seek and the points they want to make,” he added.

Star should know.

He has been teaching math in a Watertown middle school for seven years using a practice based on his research. And his work there more often informs his teaching at Harvard rather than the other way around, he said.

As a working teacher, Star knows what new teachers need to do to help students deepen their knowledge of mathematics and their appreciation for it.

“In mathematics, we’re not just interested in getting the right answer, but we’re interested in how you get the right answer,” said Star. “Creative, innovative mathematicians have a word for this, which is elegance.”

Star said he enjoys mathematics’ elegance and beauty, but what really keeps him going is teaching math and teaching in general.

He recalled that after graduating from Brown University with a degree in electrical engineering, he received two job offers, one from an engineering company and another from a private high school.

The decision was easy, he said.

“It was a sort of a fork in the road,” said Star. “I could have made three times more money as an engineer, but I ultimately decided to take the teaching job. I never looked back. … It’s not the mathematics that drives me so much, it’s the drive to help children make sense of and understand the mathematics that they are learning in school and to help them experience the very beginnings of the beauty of mathematics.”

Part of a series of occasional features marking the 250th anniversary of the Declaration of Independence

Celebrations of the 250th anniversary of the Declaration of Independence have called attention once again to the oft-repeated origin tale of how the revolutionary document sprang from Enlightenment ideas — natural rights, the social contract, and separation of powers.

But other parts of the story often get less attention. That is how the founders’ thinking was affected by existing social and political conditions on the ground, as well as their familiarity with the evolution of English constitutional history.

The independence movement really took hold when political leaders grew angry over the imposition of Parliament’s tax laws, such as the 1764 Sugar Act, the 1765 Stamp Act, and the 1767 Townshend Acts that imposed tariffs on British goods. And their efforts to oppose British rule were bolstered by early forms of self-government that had emerged in the American colonies. 

“Town meetings and colonial assemblies made the Revolution happen,” said Robert Allison, Ph.D. ’92, a Suffolk University historian who studies early American colonial history. “Parliament declared that it had the power to govern the colonies ‘in all cases whatsoever,’ but each colony already had an elected assembly which made laws for it. And in New England each town had a meeting — at least once a year, but in many cases when any 10 tax-paying adult men signed a warrant for one — that made laws and elected all officials in the town.”

As the movement took shape, the framers of the Declaration were heavily influenced by 17th century political philosophy, specifically John Locke’s ideas of natural rights and government by consent and Montesquieu’s doctrine of separation of powers.

But they had to adapt some those ideas, many of which had never been tried in practice, according to Michael Klarman, Charles Warren Professor of American Legal History at Harvard Law School.

One example is the way they took Locke’s foundational formula of universal natural rights, “life, liberty, and property,” and substituted the right to property with “the pursuit of happiness” in the Declaration of Independence.

The antecedents to the phrase “the pursuit of happiness” can be traced to James Wilson, a signer of the Declaration and the Constitution, said Danielle Allen, James Bryant Conant University Professor.

Wilson is one of the least-known founders, but a recent book, “The Lost Founder: James Wilson and The Forgotten Fight for a People’s Constitution” by journalist Jesse Wegman, sheds more light on his role in the country’s founding, Allen said.

Essentially, Wilson advanced the idea that one of the central purposes of government should be something broader and higher than the preservation of individual property.

“Wilson drew on the Scottish Enlightenment for a common-sense approach to human morality that put happiness at the center, as a concept of human flourishing and well-being,” according to political philosopher Allen. “And then John Adams picked up that line of argument and used the vocabulary of happiness to make the case for what kinds of government should be formed throughout the fall of 1775 and the spring of 1776.”

Also by the spring of 1776, the right to property had been connected to a defense of enslavement and had become a contested term, she said. “The Lockean triad was life, liberty, and property. So, life, liberty and the pursuit of happiness was keeping the issue of slavery out of that important preamble.”

As well-educated people, and as British subjects themselves, the founders were well-versed in 17th and early 18th century English history, a turbulent period during which England became a constitutional monarchy, in which the king could no longer act without Parliament’s support.

“This was part of the cultural formation of the people who made the revolution,” said Eric Nelson, Robert M. Beren Professor of Government. “These were people who considered themselves as Englishmen, and they were familiar with the entire tradition of English constitutional history, including the fearsome debates over liberty, sovereignty, monarchy, and property going back at that point nearly 600 years.”

The founders drew upon many Enlightenment intellectuals, including Locke and his influential “Two Treatises of Government” (1690). But Locke was only one of many thinkers who influenced the founders, said Nelson.

“Locke was a major figure within the traditional of political thought and constitutional refection that the founders were familiar with,” said Nelson. “He was a very important source, but it’s wrong to say that it was ‘The John Locke Show.’”

Among other political theorists of the era was James Harrington, who wrote “The Commonwealth of Oceana” (1656) on what makes an ideal republic and what makes a government stable.

Also influential were Algernon Sidney, author of “Discourses Concerning Government” (1698), a critique of royal absolutism and a defense of republicanism and popular government and the natural equality of all men; John Trenchard and Thomas Gordon, who wrote “Cato’s Letters” (1720-1723) on the importance of liberty, government limits, and the rule of law; and David Hume, who penned “A Treatise of Human Nature” (1739-1740) on liberty and checks on powers.

In addition, the 1215 Magna Carta, which set limits on royal power, and the English Bill of Rights of 1689, which limited the powers of the monarchy and established Parliament’s rule over the British Crown, were part of the intellectual origins of the founders’ vision for the new republic.

“The most important intellectual source they had was an understanding of 17th and early 18th century England, when there had been a big fight over the role of Parliament versus the role of the Crown, and they were really versed in that whole political world as though it was yesterday,” said Allison.

Even as the ideas of liberty and self-governance drew popular support to the independence movement in the colonies and abroad, there was considerable disagreement among the founders about the new kind of political system they were building.

“We spend a lot of time in our political culture asking, ‘What did the founders think about X?’” said Nelson, the government professor. “The founders were a very heterogeneous group of people who thought very different things and had very different political and constitutional preferences and ideas. So, the question ‘What did the founders think about X?’ is almost always badly posed.”

The result was a lot of compromise, improvisation, and much experimentation.

For that reason, Klarman, who wrote “The Framers’ Coup: The Making of the United States Constitution,” warns of the dangers of treating the nation’s foundational documents as “divinely inspired” and its framers as “demigods.”

“I think the framers were pretty virtuous, and they were trying to do what was good for the country, but they had interests, just like everybody has interests, and among them was to protect property rights,” said Klarman. “But we turn the founders into God-like creatures, who were ultra-virtuous statesmen who had no interests or prejudices, but simply were divining the best form of government.”

It’s an argument that prevents necessary changes in the Constitution, he said, to make institutions more responsive to these modern times and the union more perfect.

“Throughout American history, there has always been this tendency, usually a conservative-leaning tendency, to say, ‘How dare you touch the Constitution? How dare you criticize how the Senate is constructed? How dare you criticize the Electoral College?’ and that we ought to be reluctant to change any of these things because the framers were so extraordinary, so selfless, so divinely inspired … We should decide these questions on their merits, not with a heavy thumb on the scales in favor of what the founding generation believed.”

More like this

• 

  Hot off the presses: The Declaration of Independence

  A flawed first copy — dashed off in all-nighter — endures as template for democracy in action

• 

  America the beautiful

  Celebrating nation’s natural wonders through art on 250th anniversary

• 

  One of first to receive University honorary degree? George Washington.

  Archivist says it’s an early example of ‘long tradition of honoring national leaders’

• 

  How did Ben Franklin’s barber keep up with news on war for independence?

  Houghton exhibit showcases newspapers, pamphlets, other media, offering on-the-ground view of what regular citizens knew, and when they knew it 

• 

  Helping to give birth to nation — and Harvard Med

  School founder John Warren numbered among alumni who were part of revolutionary generation

• 

  Walking in Harvard’s ‘Revolutionary footsteps’

  Exhibit traces University’s role in America’s birth — from campus barracks to Founding Father alumni

• 

  Did the British unleash biological warfare against Washington’s troops?

  Historians trace role of physicians, medicine, disease during war in articles marking 250th anniversary of Declaration of Independence

• 

  High-tech for the 1700s

  Exhibit showcases tools that powered Revolutionary America

Part of a series of occasional features marking the 250th anniversary of the Declaration of Independence.

The U.S. boasts a diverse array of stunning natural landscapes, from the dense Louisiana bayous to the peaks of the Rockies. 

“Nature is our commonwealth — our common wealth,” said Horace D. Ballard, Theodore E. Stebbens Jr. Curator of American Art at the Harvard Art Museums.

Early European colonists in the Americas, Ballard explained, sent friends and colleagues plants, seeds, and the bones of mastodons, as evidence of the marvelous landscapes that produced them. Later, in the 1850s and ’60s, Americans with little ability to travel would pay to crowd into tight rooms for a chance to view a single painting of Niagara Falls. 

“When it comes to the United States, with all of its thousands of microclimates and ecosystems, it’s dizzying,” Ballard said. “One of the most profound and simple — but not naive — pronouncements we can make and hold onto on the country’s 250th anniversary is this incredible, dizzying diversity we’re all so proud of, and we all benefit from, socially and environmentally. E pluribus unum — out of many, one — defines the way the world has always seen the U.S.”

You can find many examples of artwork documenting the nation’s natural beauty at Harvard Art Museums. Ballard and other experts walk us through a selection below. 

---

---

---

---

---

---

---

---

It has become a science fiction trope come to life.

Tech giants OpenAI and Anthropic disclosed in July that versions of their AI models broke through safeguards known as a “sandbox, gained access to the internet, and hacked servers of outside companies during internal tests of their cybersecurity capabilities.

This week, AI Security Institute, a U.K. government research entity reported additional incidents, saying AI agents created fake online personas in order to improperly access real people and companies during security tests it conducted on the two tech firms.

OpenAI CEO Sam Altman characterized the July breach as an “unprecedented” and “significant security incident” caused by rogue AI agents. Its investigation has since uncovered evidence of additional breakouts. Anthropic blamed human error involving an evaluation partner.

In this edited conversation, James Mickens, Gordon McKay Professor of Computer Science at SEAS and director of the Berkman Klein Center at Harvard, explains what these breaches mean for the future of AI security.

---

Gazette: How plausible are the explanations offered by OpenAI and Anthropic?

James Mickens: It’s hard to say. I think it is very commendable that both OpenAI and Anthropic decided to release public incident reports about what happened. They could have very easily kept this to themselves and not let people know that these models have these autonomous abilities to get past these sandboxes.

However, because the public doesn’t really know the details of what happened, it’s hard to fully verify the timeline and the narrative.

The explanations put out by OpenAI and Anthropic are certainly plausible. Those public-facing stories are certainly what could have happened. Is that literally what happened? Was there anything left out? We don’t know.

We do know both companies have talked to third-party security companies to validate stuff. But we don’t really know what’s going on.

And this points to one of the tensions with this dangerous technology. One can understand why OpenAI might not want to say, “Here’s a detailed description of our sandboxing environment.”

But on the other hand, there’s this very well-known security principle, which is that sunlight disinfects. The more eyes you have on something, the more brains you have to try to find problems.

Two of the biggest players in AI report similar incidents within weeks of each other. Is that troubling?

Yes, it’s quite concerning. I would also say these are not threats that the security community and the AI safety community are just realizing were possible.

Here at Harvard, we’ve been saying for a long time that security researchers need to be thinking more critically about strengthening sandbox mechanisms and making it more difficult for models to break out of them. Last year, my research group published a paper specifically about how to create new types of software and hardware to make sandbox escapes more difficult.

Even as recently as last year, there were some people saying, “You’re guarding against a sci-fi eventuality. You’re worried about things that really are not top of mind for people using these AI systems in the real world.”

Yes, there are frontline risks that are harmful and relatively mundane, but then there are other risks that are exceptionally pernicious and quite alarming in terms of how quickly they can affect huge numbers of people in society.

For example, what would happen if one of these models tried to break into the power grid or tried to tamper with financial systems to affect the stock market? These are very real concerns.

I think with what we’ve seen over the past week or so, we should be worried about this problem more generally.

These models are very difficult to understand. There’s a whole field of AI interpretability, which seeks to understand why models do the things that they do. They’ve made some progress, but it’s still an open challenge to guarantee that a model will always act in ways that are aligned with human preferences.

So yes, these incidents are quite concerning.

Is that a concern about the technology or about the people overseeing the technology and their priorities?

Both. First and foremost, there are technical challenges in defining what alignment means, determining whether models are acting in aligned ways, and then enforcing human-defined alignment policies when models misbehave.

Those challenges are technical inasmuch as they require us to build new software and new hardware to enforce alignment policies.

However, there are also social problems of governance. Who are the humans who decide what misalignment and alignment mean? Are they mostly people who live in the U.S.? Who work for U.S. companies? Are they people who live in China, or in India, or in Brazil? Are these questions that should be answered by companies or should governments be doing this? And how do we ensure international coherence of approaches toward these things?

So, there are some very thorny technical and socio-technical issues here that are intrinsically difficult to solve and that require thoughtful responses that are at odds with the breakneck speed that companies want to move at.

Do these incidents suggest that AI models have now moved beyond human control?

I would say these incidents demonstrate that even frontier lab companies cannot guarantee model alignment in all scenarios. We’ve actually known for some time that defining aligned behavior is difficult, enforcing aligned behavior is difficult, and that models are already sufficiently complicated and sophisticated to press us on those two issues.

In other words, AI safety risks are not a theoretical problem. They haven’t been for quite some time. And it’s naive for people to think that we can afford to not think about safety risks because we want these AIs to solve cancer or things like this. We have safety problems now. We can’t wait to think about this.

Given the winner-take-all stakes in AI, could such unflattering disclosures be a kind of humblebrag by OpenAI and Anthropic to show how cutting-edge their AI models are?

Yes, it’s certainly possible.

This is why a lot of security researchers, a healthy minority, take all of these releases with a grain of salt because first of all, the public has no way of knowing whether these types of incidents have happened 100 times this year. And we’re just hearing about the recent ones now because why?

Could it be that all the companies want to race toward Artificial General Intelligence [AI capability that matches or exceeds human-level skills] and claim the AGI mantle first? AGI is less a crisply defined technical achievement and more of a “we’ve decided that we have it and now we’re going to claim it.”

And so, cynics in the security community are saying, “We bet that sandbox escapes have happened before. You just didn’t tell us. The reason you’re telling us about these sandbox escapes now is you’re trying to pave the way for this declaration of AGI unilaterally, and you’ll be able to point to this press release and say, ‘See? That was the first glimmering of our model which has AGI.’”

There’s a lot of stuff these companies put out that technical folks and policy folks don’t take fully at face value.

Is it too late to impose regulations on AI to try to prevent this from happening again?

It’s not too late to impose regulations.

Getting the right regulations in place will be tricky because this is a socio-technical problem that requires both robust technical mechanisms for sandboxing models and societal consensus on what the trade-off should be between speed of AI development and reining in potential harmful behaviors from these models. We don’t have that second part either.

Different reasonable people will have different ideas of how we should be striking those trade-offs. The important thing, though, is that we at least have to have the conversation seriously. That’s, I think, the big thing.

What types of interventions might be most effective?

There’s been a variety of mechanisms, technical, legal, and otherwise, that have been proposed for making models safer.

I think a key challenge, though, is how do we ensure that the model vendors are actually using these mechanisms? How do we get more insight into what happens when something goes wrong? How do we understand who should be held accountable for these things?

It’s important for us to not be unnecessarily restrictive. We do want to allow these companies to engage in cutting-edge research. But, at this point, the risks are getting so large and so unavoidable that it would be irresponsible for us as a society to not think about how to reckon with these things.

It’s not as if we have no levers that we can push to try to make safety better. There are a lot of people who are thinking about how do we interpret what models are doing, how do we steer them to act in ways that are more societally aligned, and so on.

There are also regulators and people in government thinking about this too, so it’s not as if there aren’t people who care about AI safety risks.

Hopefully, this past week can be a fulcrum moment where we have a series of incidents that are so clear in terms of their implication that society will have to organize around this and try to think critically about AI safety.

Excerpted from “Gerontocracy in America: How the Old Are Hoarding Power and Wealth ― and What to Do About It” by Samuel Moyn, J.D. ’01, and published by Farrar, Straus and Giroux

For two centuries, “Democracy in America,” published by the Frenchman Alexis de Tocqueville starting in 1835, has been the most famous book about the United States. It deserves to stay that way. But it might need an update.

There have always been older people. A psalm of the Bible says that, unless they die early for some reason, human beings get 70 years — 80 if they’re lucky. Genesis mentions generations of superagers at the start of humanity — like Adam, Noah, and Methuselah, the latter of whom celebrated his 969th birthday. And elders have long been accorded preferences and priorities. Whenever seniority is used to allocate positions or resources, an impulse toward gerontocracy — the empowerment of the old — is at work. Another biblical passage, this one from Proverbs, holds that “gray hair is a crown of glory.” The “preference for age and seniority,” one historian has remarked, has been “shared by all those corporate institutions which set a value on hierarchy, stability, and continuity.”

But there is something new that has supercharged elder power. Starting in Tocqueville’s 19th century, a “great aging” took place as average life expectancies increased, thanks to the reduction of childhood mortality and eventually adult disease. Today, a new crop of Methuselahs is living beyond 110 or even 120, in the case of Frenchwoman Jeanne Calment, who died in 1997 at 122, the oldest age on record. But the real change is not that more people live to 100 but that so many reach 80 or 90, even as fewer are born. The political effects of a citizenry with a higher median age have been mounting.

At the dawn of modernity, Tocqueville commented that a new political science was needed for a new age; in the 21st century, an even newer political science is needed for the new aging. 

The virtue of life extension, obviously, is that life is generally preferable to death — especially if the debilitating effects of age are also driven into the future. I remember my father, as he progressed toward the unavoidable, often counseling me never to get old; one of my mother’s nuggets of wisdom, as she has lived past 80, is that the alternative to aging is even worse. “Death, no matter how inevitable, is an abrupt cancellation of indefinitely extensive possible goods,” the philosopher Thomas Nagel writes, adding that death is no less tragic for those lucky enough to live long rather than short lives. 

But there are vices to a culture of extended life — and one of the biggest is political, in the broadest sense. An aging society is more set on preservation than on renovation. And then there’s the unfairness involved in blocking the young and hoarding their inheritance while mandating them to care for the old and requiring them to serve a government helmed by their elders.

Americans definitely worry today that our democracy is at risk — that new forms of oligarchy are a frightening possibility. But what if we have long been drifting into an unsuspected form of rule, one less apparent because it is only sometimes visible in government: a new kind of society that privileges many older Americans — to the detriment of all, including many senior citizens themselves?

Welcome to gerontocracy in America.

←→

President Donald Trump is old. His predecessor, Joe Biden, is old, too — and was even older-seeming when it counted. The evidence of Biden’s decline during the 2024 election campaign altered the course of American history, leaving a storied republic on the brink.

For years, loud voices had cautioned against even broaching the topic of Biden’s debilitation. One observer went so far as to boast that gerontocracy serves America well. Even he flip-flopped in the face of Biden’s undeniable limits. The experience of 2024 brought the political consequences of aging leadership home to millions of American citizens. Of course, Democrats continued to show a default respect for the aged — except for when they were combating Trump by highlighting his own age-related dysfunction.

Of course, not all aging politicians are decrepit. Nor is the cognitive and physical decline of politicians — a genuine problem beyond the presidency, with the median age of congresspeople spiking to over 60 and of federal judges to 68 — the main effect of aging on politics.

An aging society is more set on preservation than on renovation.

Our political system is democratic, meaning that the authorities — no matter their age — depend on electoral and other popular support to rule. And our society is capitalist, so most actual power is exercised by people outside government, especially the wealthy. To inveigh merely about the age of politicians is, in many ways, to miss the point of how power works today. If the voters and the wealthy are old, it may not even matter whether the politicians are young.

To a gobsmacking extent, seniors, with their high rates of participation, dominate elections, especially the apparently boring local and off-cycle ones. The median age of actual voters these days — meaning that equal numbers of voters are older and younger — is about 52, and on the way up. Outside presidential contests, it is 55 or 56; in primaries, it is 65. Even older funders disproportionately finance the races. The most common age of donors in recent elections can run as high as 70. Finally, older Americans are backed by the best-funded advocacy group in world history — the AARP — and enjoy the most governmental benefits by a vast degree. The government is bought and paid for by those who are up in years and is organized for their sake. Trump’s “big beautiful bill” is only the most recent exacerbation, cutting down the welfare state — except elder entitlements — to make up for lost taxes. 

It doesn’t stop there. Aging Americans control the biggest bank accounts and stock portfolios, partly as a function of living long enough to accumulate more and more wealth without giving much away. In our time, the old rich have pulled away from everyone else. In the two decades or so straddling 2000, households headed by adults older than 65 improved their median net worth by 42 percent, while the wealth of families of adults 18 to 34 fell by 68 percent. American oligarchy is, by and large, an “oldigarchy.” 

Older people are also hoarding jobs. Following the abolition of mandatory retirement in most professions, older people are choosing to stay in some of the best positions, accumulating more money and blocking pipelines. The average age of CEOs and professors alike has gone through the roof. The United States, which has the highest wage inequality by age in the world, has seen the numbers gallop: Between 1979 and 2018, the pay gap between those over 55 and under 35 has increased by 61 percent. In those same years, the proportion of older people in the workforce has also exploded: The share of workers over 55 rose a shocking 88 percent.

The elderly possess the choicest real estate in great cities. The housing crisis, now well-known, is partly a story of old people refusing to downsize as they age. The age group most likely to own a home in America, at a rate of over 80 percent, is 70 to 74, with those 75 and older second highest. Shockingly, the age of the median homebuyer leaped from barely 30 in 1981 to 53 in 2022. The failure of abundant supply is also partly the fault of the elderly, since their political mobilization arrests development and blocks new builds. In defense of their incomes, wealth, homes, or land, they have driven tax revolts so that they don’t have to contribute to the common good.

In short, it’s not just that presidents and other politicians are old. It’s not just the cognitive or bodily decline they suffer. Rather, what’s most important is that such leaders represent an aging constituency that controls the political system. They are also the visible face of the elder domination of private forms of power, especially wealth. What’s more, with millions privileged by gerontocracy, there is no way to separate the age of our elites from their ascendancy. In America today, age is the modality in which class is lived (with apologies to the great, late cultural theorist Stuart Hall, who said the same thing about race).

Today’s gerontocracy is not anybody’s malevolent plan, exactly. It is more an accidental by-product of a legitimate and understandable desire to survive, with political effects that nobody has reckoned with. Another day, month, or year among loved ones — what downside could that have? Such a question would have been rhetorical until recently, but it has now snuck up on unsuspecting societies. Boomers aren’t distinctively evil (well, some are). Rather, the fact that they are so numerous, and the fact that they are aging in an era when the extension of life is normalized, makes an endemic syndrome more obvious.

That syndrome is a gerontocratic crisis of succession on the scale of American society itself. It often involves a deathwatch for old people. The melodrama of succession — waiting for the old to make way for the new and give up effective power — is a feature not only of our politics but also of our economy and of our culture writ large.

Should Trump fail, for whatever reason, to survive beyond the midpoint of his second term, J.D. Vance would succeed him as the youngest president in all American history, beating Theodore Roosevelt, who occupied the office just shy of his 43rd birthday. Even then, American gerontocracy would remain. It would retain its grip on the rest of government, and on our society as a whole.

←→

If Americans deserve a society that is oriented to innovation and problem-solving and equips and launches the young into the prime of life, the country is heading in the wrong direction. “They live in memory rather than hope,” the Greek philosopher Aristotle observed in his ancient diatribe against the aging and in defense of people in their prime.

The presumption that people grow more conservative as they age is one of the hoariest political clichés — but it is nonetheless true. Of course, these days the far right often draws from the very young. Even so, it’s abundantly clear that age and outlook tend to correlate. Who believes the children are our future? Not American seniors. The politics of educational funding for public primary and secondary schools shows a definite connection between aging and stinginess — with voters beyond their child-raising years preferring lower taxes to decent schools. Far beyond education, elder priorities, according to experts, have come “at the expense of investments” that “promote long-term growth” such as housing and infrastructure. Enterprise in the broadest sense is not a priority of gerontocratic societies. And it has been demonstrated that older voters register more sensitivity to inflationary pressure on the continuity of their way of life than to low growth, which they do not feel affects them.

Shockingly, the age of the median homebuyer leaped from barely 30 in 1981 to 53 in 2022.

The oldest are also the most implacable opponents of immigration — even though newcomers are the likeliest to provide the care an aging American population needs. One reason for the disconnect is that older Americans incarnate the demography of the past. Ethnic and racial diversity in the United States is at its maximum in current infants and lowers all the way up the age continuum to the 85 to 99 set. The most common age for whites is 58 — more than double the most frequent age for minorities. This wide disparity, partly due to existing immigration, sets up a choice between preserving older people’s conception of the America they knew and allowing immigrants to fulfill the basic needs of older people themselves.

Most important, gerontocracies are prone to let long-term problems fester and worsen. As the individual life dwindles, playing for time in the face of impending catastrophe is a psychologically appealing stratagem of avoidance and denial. The aged preference for addressing the short term leads to the imposition of heavier and heavier consequences for the future — as those understandably concerned about an indebted nation or a warming planet stress.

A fundamental imperative is therefore at stake. Americans must transcend gerontocracy if we are to realize our collective aspiration to innovation, as well as achieve other important goals such as intergenerational equity, fair political representation, and even an existential acceptance of limits as we age. Everyone likes to think of themselves as ageless, but we are finite, with just one passing life to live. Ultimately, gerontocracy is built on a denial of this transience. The aging often respond to mortality by holding on to power as long as they can — but if life is inevitably lived in stages, then succession doesn’t need to be as melodramatic or unfair as it is now.

←→

If modern gerontocracy functions through a new political environment, it follows that a new range of solutions is required. Age limits for political office are a must. Other fixes range from amplifying the political voice of younger voters and disrupting the stranglehold of the old people’s lobby on our politics to policies that facilitate intergenerational wealth transfer. The schemes I offer seek to advance “intergenerational equity.” But unlike the most vocal partisans of the cause, I don’t oppose the American entitlement state for seniors. Such voices insist that an end or reduction of perquisites like Social Security must occur in the name of fairness. The truth is the opposite: only an expansion of our generosity, including the promise of long-term care, can overcome the iron grip of old people on power.

While gerontocracy is hardly the only source of our difficulties, no one interested in how injustice and social strains are shaped and reproduced in America today should ignore how deep it goes. Gerontocracy has to be assessed properly, for the sake of a new intergenerational political compact. In the end, taking it seriously is essential for treating everyone appropriately and fairly.

There is still a chance for a reset. The extension of the lifespan has had political consequences, many for the worse. It has already brought an ancient phenomenon — the authority of elders over everyone else — back in a dangerous new guise. Biden and Trump have exposed one part of our gerontocracy. Ripping away the curtain that hides the rest can prepare us to dismantle the system and create something new.

Copyright © 2026 by Samuel Moyn. All rights reserved.

George Usmanov did some unforgettable sightseeing this summer.

Over 25 days in June and July, he passed by ghost towns from both the 19th and 20th centuries and acres of forest, “completely charred.” He saw a river run yellow out of a Superfund site and the sun rise purple, thanks to a mix of “desert dust, mist from the Rockies, smoke that was coming in from wildfires.”

Typically, though, he had no one with whom to share those moments. Usmanov was taking part in the Tour Divide, widely regarded as the world’s most difficult bike race. It stretches over 2,700 mostly unpaved miles from Banff in southern Alberta, Canada, to the U.S.-Mexico border. And during his biking days, “about 70 percent of the time, I was entirely alone,” he said. “Even if you’re only separated from other racers by a couple of miles, the trails are extremely twisty and you don’t see them.”

This was in part a working holiday for Usmanov — and that kind of isolation was his object of investigation.

The Harvard Medical School research fellow in neurology at Massachusetts General Hospital is engaged in postdoctoral research on stroke — the nation’s fourth-most-common cause of death, and a condition in which every minute counts. The distances he covered are the same ones that separate some of the continental U.S.’s most isolated communities from centers of effective treatment.

Usmanov, 28, is not a doctor but a medical sociologist. His studies focus on “illness experiences”: how felt symptoms, social connections, and cultural norms encounter and shape treatment.

Stroke — with its various forms and subtle symptoms — is of obvious interest. Most stroke patients don’t feel pain. Instead, at onset, they experience “numbness, tingling, or some sense of disruption: you’re driving a car and suddenly you can’t turn the steering wheel. Or you’re using the TV remote and suddenly you don’t know how to change the channel,” Usmanov said.

Meanwhile, though, a clock has started ticking. “From that moment, parts of your brain are dying, at a rate of around 1.9 million neurons a minute,” Usmanov said.

In cases of acute stroke, the field of neurology has come to see the first 60 minutes after onset as the “golden hour”: If patients are examined and treated with an intravenous clot-busting drug like alteplase within that window, their odds of a full recovery improve dramatically. And clot-busting drugs can be used effectively for up to 4½ hours after a stroke’s onset.

The trouble, Usmanov said, is that “in the United States, the vast majority of people don’t get to a hospital in time, meaning they’re missing out on the most effective treatment.” (Mechanical thrombectomy — using a stent or vacuum to remove a clot physically — can be used up to 24 hours after onset, but it’s more invasive.)

To be sure, even most acute strokes are not lethal. But up to one-quarter of acute stroke survivors die within a year of onset, and survival and recovery rates vary greatly by stroke type, patient age, and the speed and type of treatment. In general, Usmanov endorses what has become a neurological maxim: that in cases of stroke, “time is brain” — meaning there’s an immense public health interest in shortening the gap between onset and appropriate treatment.

Usmanov noted that this country has made important strides toward making “golden hour” arrivals more likely — at least in theory. Individual medical centers have worked to shorten “door-to-needle” time: from intake to radiology and infusion. And not only can ambulances or helicopters shorten trips from homes to “stroke-capable hospitals,” there are more of those than ever before.

A team of Harvard researchers found that in 2019, 91 percent of Americans lived within one hour (by ambulance) of a center that could deliver alteplase; in 2011, it was just 80 percent. Further, they found an additional five percent can be reached by “telestroke”: a remote examination over video that has proven effective in lowering mortality and improving outcomes.

But one large region has remained stubbornly above the national average — the Mountain West, the “exact corridor” through which the Tour Divide passes, Usmanov said.

Usmanov has been an avid biker since his youth in Minneapolis. And this isn’t the first time he’s used his bike as a platform for rolling ethnography. During his graduate studies at Georgia State University, he took a similar ride to visit the informal communities that arise at campsites and RV parks across the Deep South.

“A bike is just an extremely useful and effective way to witness space, I think,” he said from his office in Somerville.

But in the Rockies, he found community harder to come by. There are fundamental problems of logistics — long distances, snowed-over mountain passes closed each year, unmaintained roads even in summer. Isolation is the rule even when ties are close. “The cost of seeing each other is so high,” he said. “Your daughter could live on the ranch next door, and that could still be an hour away.”

Moreover, the videoconferencing solutions behind telestroke — and much of post-COVID work and socializing — are not yet adapted to the heights Usmanov traveled. For about two-thirds of his ride, his phone had no connection at all, and simple sites could take minutes to load even when service was available.

Again — even with the benefit of sick leave, broadband internet, modern roads, and denser neighborhoods — Americans still tend to begin treatment too long after a stroke’s onset. So Usmanov has some justification for arguing that we need to focus on the first phase of stroke’s unforgiving timeline: between onset and calling 911.

For years, the public health establishment has tried education paradigms. Since 2013, they’ve used acronyms like F.A.S.T. to help laypeople recognize the telltale signs of stroke. But further study found that 30 days after an educational video, many participants had forgotten the symptoms, and that rates of timely 911 calling declined as well.

And in the Rockies, as in many lower-income precincts, there’s an element of class at play. “Generally, I met people who have done highly laborious work for years: construction, ranching, in the service industry,” Usmanov said. “And they tended to tell me similar things: ‘I’ve been in pain my whole life.’ ‘I broke something, but I had to work through it.’ Or ‘I experience weird sensations as part of my job: tingling, dizziness, numbness.’ Those are just the demands of the job — but they’re also symptoms of stroke.”

In January, Usmanov was the lead author of an American Heart Association paper finding that patients who were with family or trusted bystanders at onset tended to arrive at hospitals earlier than those who weren’t — or even than those who were with large crowds of strangers.

And his Great Divide ride underscored how much community matters in unforgiving terrain. 

Some nights he’d camp serendipitously with other racers, for camaraderie as much as for safety. He made a “good friend” in one fellow racer, only to watch him drop out after developing nerve damage: the consequence of biking over bumpy dirt track with inadequate suspension.

And 500 miles from the finish line, Usmanov was contemplating dropping out after a crash and without water. “A rider named Bob passed me, gave me a little Gatorade, a kind of a pep talk, and went on his way.” (Only about half of the roughly 260 racers completed, with Usmanov middle of the pack.)

Usmanov’s ride had one last goal in mind: it served as a fundraiser for the Global Brain Care Coalition, a nonprofit that shares his ambition to find preventative care for neurological ailments. (It has raised $20,000 at time of publication.) And the Rockies taught him that there may be promising interventions to be made between individual vigilance and institutional change.

“Spouses, neighbors, church members can take blood pressure together, can discuss health behaviors — and that will do a lot more than this ad hoc approach, of one individual at a time,” Usmanov said. “We’re social creatures, and that’s where we have to target change.”

It’s a fitting takeaway for a sociologist after a long, lonely ride through the mountains.

The question is no longer whether young people are using AI for mental health care, but the soundness of the advice they receive, health officials say.

A recent survey showed that the use of AI chatbots for mental health care among adolescents and young adults has increased 60 percent in just a year, a trend that survey organizers said is likely to continue.

Hao Yu, Harvard Medical School associate professor of population medicine at the Harvard Pilgrim Healthcare Institute, led the study, which had 1,727 participants.

Yu said as AI helps fill a treatment gap left by a rising problem among young people in the U.S. amid a shortage of mental health professionals, the focus should now turn to ensuring the counsel being offered is helpful and not harmful. The interview has been edited for clarity and length.

---

Gazette: What did your survey find?

Hao Yu: We did a similar study last year, and at that time we found one in eight young people, between the ages of 12 and 21, use AI for mental health problems. This year we found an even higher proportion, almost one in five. That means nearly 20 percent of young people in this country are using AI for mental health advice. That’s a rapid increase in one year.

Not surprisingly, we found variation in AI use across youth groups, with relatively older youth more likely to be using AI, and females more likely to use AI for mental health issues.

It may surprise some that we found the majority of youth using AI did not disclose their use to any adults, either parents or physicians. That might be concerning, because there is misinformation available through different AI models, which are not designed to provide mental health advice.

So, we would encourage parents and physicians to talk with their youth regarding their potential use of AI. They may be able to identify if their child has a mental health issue or help their child distinguish between accurate information and the misinformation we know is available online.

What does this tell us about the state of young adult and adolescent mental health care today?

This definitely shows that there’s a big issue facing the U.S. today. On one side, we have a rapid adoption of AI across different parts of our society. On the other, there is an increased prevalence of mental health issues among young people. Putting these two together makes it not so surprising when we see that youth are increasingly turning to AI for mental health advice.

Does it also highlight that there’s not enough capacity in our mental health care system?

Our healthcare system has an acute shortage of psychiatrists, especially child psychiatrists, and only half of those with mental health needs actually receive mental health treatment.

So, if a youth has mental health issues, AI offers some advantages. AI is patient. AI is available anytime, 24/7. You don’t need an appointment to talk with AI, and AI is nice. It’s rare for AI to be harsh or become impatient. So there are advantages, especially for people living in rural or remote areas with severe mental health workforce shortages.

How much do we know about the quality of the responses that come from these general-use AI chatbots? Is it still a concern that that kids and young adults may get bad, even harmful advice?

That’s still a concern. And it’s true that tragedies like suicide attempts have happened when youth talked with AI about mental health issues. AI companies promised they would install guardrails but there’s a lack of empirical evidence about how those promises are being kept.

Beyond this specific study, many researchers are interested in assessing the safety and efficacy of mental health advice offered by AI. More people are paying attention, so we’d hope there would be some improvement, but that remains to be seen.

The paper mentions that these AI models are designed to be friendly, agreeable, and flattering. Might kids also be getting information that is reinforcing in a positive, but still unhelpful way?

Most AI models try to be appealing, flattering, as a way to keep their customers, and many people, especially youth, easily hooked. But it’s still a concern that the advice, the information provided by AI, be really useful.

In our study, we found that the vast majority of youth, over 90 percent, say the advice is useful, but that’s still an empirical question we need to assess: Is AI-provided mental health advice really helpful and accurate?

The safety and efficacy of AI models is a priority for research because, as we said in the beginning, these things are already happening, and the trend among youth is increasing. As a society, we have the responsibility to make sure that what they are receiving is accurate and helpful.

You mentioned that there are AI chatbots that are designed to be therapeutic. I’ve seen earlier studies that show that they can be very good, but their development has been slow. Is a key message that we need to accelerate the development of these therapeutically trained bots?

Yes. We have an acute shortage of child psychiatrists who can meet unmet needs among our youth, so we want AI models that can mitigate the shortage. It’s not surprising, though, to see people concerned about the potential quality and safety implications of even the specifically trained, mental-health-oriented AI.

That’s why some policymakers, as in Illinois, have taken actions. Ultimately this is an issue of patient safety.

Where do you see this trend going? We’ve seen an increase from one in eight adolescents and young adults using AI for mental health needs to one in five. Are you planning on doing the survey next year, and do you expect it to be an even greater percentage?

We definitely will have follow-up surveys, and most likely will see an increased number. It’s not surprising given that AI is now part of daily life for youth, and that there are unmet mental health needs. AI is readily available and endlessly patient.

Were there any surprises in this for you?

One surprise is that we didn’t find a significant difference between rural and metropolitan areas, even though we know that a rural-urban digital divide exists and that rural and remote areas have a severe shortage of child psychiatrists.

If confirmed by future research that rural and urban kids are equally likely to use AI, that might help reduce the adverse effects of a workforce shortage in rural areas. AI has that potential for sure. But we want to make sure these specific AI models meet clinical standards for mental health advice.

The research described in this interview received funding from the National Institute of Mental Health.

Seventy-nine percent of Gen Z want a job mainly focused on helping people or making a positive difference in others’ lives. But only about half of workers in that 18-29 age bracket who want such a position say they have one. 

Those are the findings of a new report from the Harvard Graduate School of Education’s Making Caring Common project, in collaboration with Gallup and the Walton Family Foundation. The survey examines what members of Gen Z want from their work and how it affects their emotional well-being.

The data, taken from a representative sample of 2,436 people born 1997-2012, goes against popular stereotypes, according to Richard Weissbourd, faculty director of Making Caring Common and a senior lecturer at HGSE. 

“Gen Z is often characterized — in my mind, cartooned — as selfish and lazy, as not wanting to really invest in anything deeply. So I was really surprised, partly because of this misconception, to see that 79 percent of our respondents report that they wanted to have a job in the future that was mainly focused on helping other people,” Weissbourd said. “It was just a very high percentage. And it’s not just talk. Many Gen Zers are taking concrete actions based on their desire to do good, according to Deloitte surveys, such as choosing companies based on their social impact.”

Weissbourd explains why helping Gen Z find meaningful work could have a big impact both on the culture and in the lives of the young adults — and why there may be some good news on the horizon. The interview has been edited for clarity and length.

---

Gazette: What kinds of jobs do Gen Z find meaningful? 

Richard Weissbourd: We asked two different questions: whether they found their jobs meaningful, and whether their job was primarily focused on helping other people. 

Among Gen Zers who told us they found their jobs meaningful, about twice as many (84 percent) worked in jobs focused on helping people — such as jobs in healthcare, education, and the military, as those who did not work in these kinds of jobs (44 percent).

In response to an open-ended question about where they find meaning at work, we also found that many Gen Zers focused on those aspects of their jobs that helped others, such as a dishwasher who finds meaning in playing an important role in providing people food or a retail sales worker who finds meaning in helping people find clothes that make them feel good about themselves.

So I think two things are true. One is that a lot of young people want a job that’s directly helping others, like a job in healthcare or social work or teaching. But also, many young people are finding meaning in the aspects of their jobs that involve helping others, even if it might not be as obvious. 

This research only spoke to Gen Zers. So this can’t really tell us how Gen Z compares to other generations, is that right? 

That’s correct. From this and other data, there isn’t clear evidence that this generation is more altruistic or more socially active than other generations in general, but there certainly isn’t clear data showing that they’re less altruistic or less socially active. 

What’s different among Gen Zers who report feeling purpose and meaning in their lives? 

Over half of Gen Zers report mild (30 percent), moderate (18 percent), or severe (10 percent) anxiety and depression, and a whopping 70 percent of Gen Zers over 18 report mild to severe anxiety and depression. But those who say they have felt a sense of meaning or purpose in their lives in the past month are much less likely to report symptoms of anxiety and depression than those who do not feel their life has felt meaningful or purposeful. 

Meaning and purpose are the deep engines of our lives. They provide this deep gratification that’s beyond hedonism, beyond temporary joy. We can talk about banning cellphones in schools, which I think is a good thing, and providing more therapists, which I also think is a good thing, but my guess is that helping other people and making a positive difference in others’ lives is significantly more likely to strengthen Gen Zers’ mental health than those other strategies.

You found that just over half of Gen Z adults (56 percent) who want a job mainly focused on helping people have one. When asked about perceived barriers to taking those jobs, many of them said they perceived care-oriented jobs to be low-paying and emotionally draining. Are they right? 

They are right in the sense that a lot of these jobs aren’t affordable. I mean, if you want to be a childcare worker or preschool provider in a city like Boston or New York, it’s very tough. We need to figure out ways to pay people more, reduce housing costs, provide loan forgiveness for people in helping professions in under-resourced areas, et cetera.

Certainly there are helping jobs in ed tech, philanthropy, and other industries that pay well. But we should also be creating-less expensive pathways into careers in healthcare and education. 

The emotionally draining aspect is interesting to me. These jobs can certainly be emotionally draining, yet in our data, when people feel they’re making a positive difference in other peoples’ lives, they also report less anxiety and depression. Teaching, for example, can be very depleting, but it’s also very gratifying for a lot of people. Both of these are true. 

So is it fair to say that Gen Zers see helping jobs as a nice-to-have in principle, but only as long as they can pay their bills? 

It’s possible. Here’s one finding that speaks to that question. We asked people: “You have a job that provides you with a feeling of meaning and purpose. You’re offered a new job that has double the salary but will have significantly less meaning and purpose. Would you take it?” 

To summarize, people were roughly split: 46 percent said they would take the higher-paying job, 28 said they would keep the meaningful job, and 26 said it would depend on the specifics. 

Now, when we dug a little deeper, we found that perceived financial security clearly mattered. No matter their actual income, adult Gen Zers who felt financially insecure were more likely to take the higher-paying job (53 percent) compared to those who were living very well (41 percent). But a significant number of people with low financial security are also taking up helping jobs. So it’s a complicated picture.

With an aging population in need of care and potential AI disruptions to knowledge-economy jobs, is this a rare moment where this population’s desire for helping jobs could align with where the economy is headed? 

Yes. I’m not an expert in the economy, but I can say a very common view is that we’re shifting from an economy that values information and knowledge to an economy that values interpersonal skills, social-emotional skills, and the capacity to care for other people.

In healthcare, for example, you can imagine AI taking over a lot of the knowledge functions that a hospital might rely on, freeing up more people to do really high-quality patient care. The gist of it is that this is a time when what young people are interested in and what the economy demands may rhyme.

Economists and Wall Street analysts frequently warn about what sounds like an immutable law of economic gravity: Growth that gets too hot for too long risks provoking an inescapable cool-down known as a recession.

But that notion is wrong, according to a new book about the history of recessions in the U.S. and U.K. dating back to 1700, “Recession: The Real Reason Economies Shrink and What to Do About It.”

Contrary to popular thinking, these “busts” are not the inevitable result of flaws in the economic “booms” that immediately precede them, and they’re not cyclical, or necessary, course correctives. In fact, recessions have relatively little effect on the economy over time compared to periods of growth, says the book’s author, Tyler Goodspeed ’08, M.A. ’11, Ph.D. ’14.

Economic historian Goodspeed discusses what triggers recessions and why war is the most prolific “serial killer” of economic expansions over the last 400 years. He served as acting chair of the White House Council of Economic Advisers during President Trump’s first term and is now chief economist at ExxonMobil. The interview has been edited for clarity and length.

---

Gazette: What’s the definition of a recession?

Tyler Goodspeed: I adhere to the definition used by the National Bureau of Economic Research, which is a broad-based economic contraction that’s spread across sectors and that lasts for more than a few months.

I think armed only with a graph of the unemployment rate, one could do a pretty good job picking out historical recessions, which are typically characterized by the substitution of a sudden, sharp upward movement in the unemployment rate for a more gradual downward or sideways movement.

You say the conventional wisdom around why recessions occur is wrong. How so?

The stories we tend to tell ourselves about recessions is that there was something fundamentally wrong in the expansions that preceded them. There was some error, some excess, some imbalance, to which recession was the inevitable, maybe even necessary, outcome.

Fortunately, that view yields several empirically testable and falsifiable hypotheses.

First of all, if there was something wrong in a so-called boom to which the bust was the inevitable remedy, then we should expect expansions to die of old age. So the longer an economic expansion has been going on, the more those errors and excesses and imbalances accumulate, the likelier it is to die.

We would expect that higher, faster, longer economic expansions should be followed by deeper, faster, longer recessions.

We should expect that the frequency of recessions should be roughly constant across time and space.

We should expect that the economy should look fundamentally different at the end of an economic recession and subsequent recovery from how it would have looked in the absence of the recovery of the recession.

We should also expect that the increased role of the state should have attenuated the depth and duration of recessions, and we should also expect that recessions should be in some sense predictable, that there should be some indicators in the height, the speed, excess, whatever, in an economic expansion that should predict recession.

In reality, when I analyzed these 132 recessions going back to 1700, all of those hypotheses are rejected by the data. All of them. Expansions don’t die of old age; deeper, faster, longer recessions don’t follow higher, faster, longer expansions; recessions have become less frequent over time.

The U.K. has long been less recession-prone than the U.S.

Typically, economies look pretty similar at the end of a recovery to how they would have looked in the absence of a recession; and recession depth and duration have been remarkably constant over time.

And finally, they’re not predictable.

So, all those hypotheses generated by the boom/bust view are just rejected by the data.

Do recessions share any common patterns historically?

There are some things that are roughly constant across time and across space. The depth and duration of recessions have been pretty constant over time. Most recessions are over in about a year; the vast majority in fewer than two.

A key feature of recessions is a sudden sharp upward movement in the unemployment rate, and that’s typically driven in the first instance not by a sudden increase in the rate at which firms lay off workers, but rather a sudden collapse in the rate at which they hire workers.

Are they in any way avoidable or inevitable?

Recessions will continue to happen because history will continue to happen. There’s never been an immortal economic expansion.

One of the optimistic findings in the book is that recessions have become less frequent over time in both the U.S. and the U.K. That is a long-run structural trend going all the way back to 1700 with no clear break at any specific points in time toward longer-lived economic expansions.

What’s driving that is households, businesses, and consumers have been learning how to better absorb the kinds of shocks that historically would have generated recessions. Two examples: Our banking systems are generally more resilient, more diversified today than they were in the 19th century, and our energy systems have become a lot more diversified, both across fuel types and within fuel types.

You argue there are three categories of “shocks” that typically precede recessions. You call them “Acts of God,” like pandemics or severe weather that slows commerce or damages crops; “Acts of Man,” such as major bank fraud; and “Acts of Church,” which are state actions like imposing consumer credit controls. But you say there’s one thing that’s been “a prolific serial killer of economic expansions” over the last 400 years: war. Why is that and is that true for all armed conflicts?

Not all wars are created equal. What I’m talking about is either world war or war that visits your soil in a profound way.

Take some of the longest recessions in history for the U.K. and the U.S. The U.K. had a very protracted recession that began in 1943 during World War II and continued all the way till 1947. They also had a very long recession at the end of and in the immediate aftermath of World War I. Another very long recession during the Seven Years War in the 18th century.

In the U.S., we had a Depression-magnitude recession during the American Revolution, and in the immediate aftermath of the American Revolution during a very volatile period.

One of the most fascinating recessions I found looking back to 1700 was in the North American colonies from 1717 to 1720. This was the peak of the so-called Golden Age of Atlantic piracy.

You had the end of a major European war during which the British government had paid private sailors to become privateers. So, at the end of the war, there were all these unemployed sailors, and they turn to piracy.

Trade and commerce in the American colonies just ground to a complete halt. It was only after some of the most infamous pirate names were killed or captured or accepted the king’s pardon that you had a resumption of normal commercial flows.

Why is there always so much fretting currently over the possibility of a recession, given that they tend to be relatively brief and less frequent? It’s especially puzzling since the U.S. notched two of its longest economic expansions ever in the last few decades, and expansions exert greater overall influence on an economy.

There’s a lot of fretting because our minds don’t deal well with randomness, and at the end of the day recessionary shocks are random. So, we’re constantly searching for predictive patterns to recessions and embed those perceived patterns in parables of boom and bust.

But the moral, boom-bust stories we tell ourselves about recessions are not entirely innocuous because if it’s the case, as I find in the book, that economic expansions die healthy and innocent, then there is a risk that we as policymakers are perpetually seeking to sedate or otherwise overmedicate economic expansions that fundamentally die healthy and innocent.

And typically, a few years on from a recession, economies look pretty darn similar to how they would have looked had they continued uninterrupted along a long-run trend.

You touch upon a really important observation, which is that the majority of years in which economies expand matters more for long-run prosperity and indeed human flourishing. And so, we probably ought to be at least as concerned about raising that rate of trend growth as we are about avoiding recessions, because ultimately the periods of expansion matter more.

Despite recent “shocks” to the U.S. economy like tariffs, the war with Iran, the spike in fuel and energy prices, and workforce reductions attributed to AI, the country is not in a recession. What signs from history do you look for to help forecast whether one is on the horizon and how detrimental it might be?

Something that I learned is that it takes either a really big shock like a massive pandemic or a cluster of smaller shocks to push a large, diversified economy into a period of outright contraction. It takes a lot.

In terms of what I look for, recessions are typically characterized by a sudden sharp increase in the civilian unemployment rate. That’s the best real-time indicator of whether you’re in a recession. It can’t predict whether you’re going to be in a recession, but it’s a good indicator of whether you are.

I find that recessions are unforecastable. They’re unpredictable. However, what the book does equip one to do is to get a sense of the kinds of shocks that historically have contributed to recession, and then the reader can ask themselves, do I think the conditional probability is higher or lower than the unconditional probability? And in the past century, there’s been, in any given year, about a 1-in-100 chance of a pandemic-related recession; there’s been about a 1-in-10 chance in any given year of an energy-related recession in the U.S.

When exploring different sides of an issue, a devil’s advocate can play-act an opposing perspective, but the exercise has limits. It’s better to talk to someone who genuinely disagrees with you.

“You actually have to make the effort to get people in the room who deeply believe in, think about, and live by a certain set of aspirations or ideals,” said Jenna Silber Storey, co-director of the Center for the Future of the American University at the American Enterprise Institute, a conservative think tank. When faculty members discuss intellectual diversity on college campuses, it’s not enough, she said, to “theoretically entertain what political conservatives might think.”

Dean Tomiko Brown-Nagin of the Harvard Radcliffe Institute agrees. When she convened a workshop on intellectual diversity last November, she and her co-leads, Harvard Ed School historian Julie Reuben and Keith Whittington of Yale Law School, took care to choose scholars and leaders whose wide range of political and ideological perspectives would put their commitment to collaboration to the test.

The group proved up to the challenge, in May publishing a white paper Brown-Nagin described as “a true negotiated document that reflected consensus.”

Storey, one of the 15 authors of the paper, credited the success of the effort in part to Radcliffe’s insistence on a truly diverse mix of thinkers.

“Simply the fact that the Harvard Radcliffe Institute invited a significant number of political conservatives to discuss this question was astonishing to many people on the political right,” she said. “It just signals right off the bat: We’re going to take your opinion seriously; we’re inviting you to be equal partners in drafting this statement.”

In the paper, “Intellectual Diversity and the Mission of Higher Education,” the authors came to several shared conclusions. They agreed that universities are facing a legitimacy crisis, in part due to public skepticism of higher education’s civic value. The tendency of faculty and students to skew more liberal than the general population is one important contributor to that skepticism, the authors found.

More concerning, “students and faculty do not feel fully free to express themselves on certain issues,” the white paper states. The group advocated for programs on campus to replace a “censorious culture with one that is more open to debate and disagreement,” adding: “We are largely united in the view that most colleges and universities simply do need more ideas and methodologies often associated with political conservatism in order to achieve their mission.”

While the group discussed areas from research to governance, they agreed that they didn’t have as much information as they needed to both diagnose and solve problems. “In the absence of rigorous evidence, it is exceedingly difficult to define precisely the problem we face and to make progress toward solving it,” the document says.

As an example of the research that would help, the authors highlighted surveys and analyses performed by workshop participants Neil Gross, a sociologist at Colby College, and Jon A. Shields, a government professor at Claremont McKenna College, and called for more study of topics such as course content and teacher and student experience.

As an outgrowth of the white paper, Brown-Nagin said that Radcliffe will soon announce a research project that will include surveys of faculty and students to clarify “the extent of political skewing on university faculties, why that might happen, and most important what, if anything, it means for classroom dynamics.”

The group also used the paper to highlight where they could not come to an agreement. Though they reached consensus on wanting a wider range of perspectives, they disagreed about whether universities should hire directly for ideological diversity.

Brown-Nagin, for one, questioned the idea that a faculty member’s political orientation inherently deprives students of a full range of ideas. Part of what people have attacked as politically skewed teaching, she said, is likely ineffective teaching — “faculty not having the tools to actually facilitate a conversation about a hot-button topic in class.”

“If what passes as political indoctrination is just a matter of faculty not having the right teaching tools, we can give them additional tools so that they can teach more effectively,” she said. At the Bok Center, she noted, Harvard offers courses on promoting healthy classroom dialogue, which are now mandatory for teaching fellows in the Faculty of Arts and Sciences.

Along the same lines, students’ reticence to speak on controversial issues might not be the result of faculty pressure. The 2024 report from Harvard’s Open Inquiry and Constructive Dialogue working group, co-led by Brown-Nagin, “found that fear of student judgment, rather than faculty retribution, is the primary motivation for most self-censorship,” she said.

“It’s important both to train faculty and to cultivate the right mindset and teach the right skills to students,” she added. She hopes that some of the new educational modules for incoming students across the University can help prepare them for deeper classroom discussions.

Reuben stressed that, while contemporary debates about academic freedom criticize universities as overly liberal, the ideological makeup of academia has shifted over time. For much of the 20th century, she said, “there was, if anything, less tolerance for what we might consider the progressive end of the spectrum than the conservative end of the spectrum.” It took decades, beginning in the 1960s, for a more progressive faculty to ascend.

Brown-Nagin expanded on the same point.

“Concerns about political indoctrination stretch far back into our history,” she said, from promoting certain narratives about slavery and the Civil War in the early 1900s to McCarthyism decades later. “There have long been fears that faculty are teaching the wrong subjects, that these politics are invading the classroom. And these concerns span the political spectrum.” It’s important, she said, to appreciate that the pendulum swings in both political directions — and that that pattern will likely continue over time.

As conversations about intellectual diversity continue, Reuben hopes that people pay attention to the paper’s emphasis on promoting professional norms among faculty. As a historian, she draws inspiration from the American Association of University Professors’ 1915 Declaration of Principles on Academic Freedom and Academic Tenure.

The white paper quotes the principles directly:

If this profession should prove itself unwilling to purge its ranks of the incompetent and the unworthy, or to prevent the freedom which it claims in the name of science from being used as a shelter for inefficiency, for superficiality, or for uncritical and intemperate partisanship, it is certain that the task will be performed by others — by others who lack certain essential qualifications for performing it, and whose action is sure to breed suspicions and recurrent controversies deeply injurious to the internal order and the public standing of universities.

In other words, Reuben said, faculty need to “step up and take responsibility for these things” — or others will do it for them.

Artificial intelligence is poised to soon change virtually every part of our lives, including the way we lead.

“There is no policy area, there is no function in the public or social sector, that isn’t going to be massively impacted by the accelerating power of generative AI,” said Jeremy Weinstein, dean of Harvard Kennedy School. “We need our curricular offerings to rise to meet that challenge.”

HKS recently unveiled a series of new initiatives across a range of areas, including financial aid and several AI and technology-focused shifts as part of a broad, 10-year strategic blueprint for the School.

As part of the plan, HKS will launch a new tech policy concentration for its master’s in public policy candidates and has hired two AI and policy professors, including MIT’s Stephen Casper, who joined HKS this summer. Casper will teach methods courses and electives on AI governance. The School also plans to look at offering a tech policy degree.

Some changes are already underway. This spring, students were offered two semester-long courses, one on the national security implications of AI led by Jake Sullivan, former national security adviser to President Joe Biden, and another on AI and policymaking that Weinstein co-taught with Sharad Goel, an applied mathematician and computer scientist at HKS.

An important focus involves getting the students more comfortable with the technology by learning how to use AI assistants to code and conduct rigorous analytics.

The goal is to ensure that they can use the powerful tools to solve public policy problems “and not just be observers or passive consumers” of AI, said Weinstein, Don K. Price Professor of Public Policy at HKS and a government professor at the College. “And that is just getting easier and easier because of the skill of the tools themselves.”

AI has “no clear analogue in history,” said Sullivan, now the Kissinger Professor of the Practice of Statecraft and World Order at HKS. Unlike other national security technologies, such as the internet or the nuclear bomb, AI development is not being led by government, but by the private sector. That makes this moment an especially “exciting” time of “radical uncertainty” for those entering both the public and private sector.

Weinstein and Goel’s class, “Public Problem Solving with Generative AI: Challenges and Opportunities,” drew inspiration from a multidisciplinary course that Weinstein, who came to Harvard in 2024, had taught at Stanford.

Combining case study and discussion with hands-on assignments, the course objective is to help HKS students, many of whom were not STEM concentrators in college, understand the technical underpinnings of AI while confronting the big policy questions this powerful new technology raises across domains from national security and science to education and healthcare, said Goel, who developed the School’s AI module two years ago.

So far, HKS students have responded enthusiastically.

Given AI’s importance, it’s essential to not only understand how this technology is being deployed globally but also how AI “works under the hood,” said Jacob Gordon, M.C./M.P.A. ’26, who took Weinstein and Goel’s course and valued the opportunity to vibe code and use Python as part of class assignments.

Gordon, who hopes to start an AI safety nonprofit, supports the push to expand and deepen the School’s focus on AI and technology.

“It’s vital, and HKS needs to move fast if it wants to contribute to this small decision-making window as the world figures out what path to take on the future of AI,” he said.

In one lab, students in Weinstein and Goel’s course tackled the challenge of getting 911 callers quickly routed to the appropriate agency or service provider when the situation doesn’t require law enforcement.

Some emergencies, like a behavioral health crisis, may demand other expertise, and dispatching armed police officers to a scene where they’re not needed may add danger to volatile situations, said Goel.

But preparing 911 call takers to quickly and accurately assess the type of help to dispatch often requires two or more years of training, a costly process that can strain municipal budgets, he added.

So, using generative AI, students built voice-training tools that used real-world data to simulate 911 callers.

The simulators tested call takers’ decision-making skills and later assessed their performance, representing a potential low-cost option that can better identify training deficits and get new call takers online faster.

“Few classes give you the chance to discuss policy issues with someone who was directly involved in shaping those policies at the highest levels of government just months earlier.”

Sarah Zheng, M.P.P. ’26

Globally, AI has already begun to remake major policy and geopolitical questions involving national security and AI regulation, said Sullivan, who served in the Obama administration as an adviser to then-Vice President Joe Biden and deputy chief of staff to Secretary of State Hillary Clinton.

“The developments of this technology are shaping relations among nations; they are shaping the national development strategies and growth strategies of nations; they’re shaping the power balances among nations” and affecting “every aspect of national security, economic security, and social interaction,” he said.

His class focuses on how nations, particularly the U.S. and China, are trying to take advantage of AI’s potential while managing its risks.

Sullivan said during his time in Washington, few who were not working on the technology side of government focused on how AI might reshape their policy domain, let alone pursued training in how to use it, but that’s no longer the case.

“You don’t need everyone to be a technical expert, but I do need everybody to have a framework and a sense of priority for this because it’s going to be so central to the big questions,” he said.

Sarah Zheng, M.P.P. ’26, who reported on Chinese technology and geopolitics for Bloomberg News and the South China Morning Post, said Sullivan’s class offered firsthand insight into how major decisions on AI and national security are being made and presented a full spectrum of perspectives on AI, from skepticism to optimism.

“Few classes give you the chance to discuss policy issues, such as technology export controls, with someone who was directly involved in shaping those policies at the highest levels of government just months earlier,” she said.

Weinstein said when HKS first began to incorporate technology into the M.P.P. curriculum several years ago, the intent was to prompt students to think about “everything from how is technology used in the defense sector; how do we think about cyber issues and cyber threats; how do we think about civic technology … and how do we govern in a digital age.”

Now, the School’s growing emphasis on AI and technology reflects the inevitable reality as well as the potential promise and peril of what they may bring.

“For me, it’s very simple. One of the most important things that we can do as a school over the next generation is to populate public institutions, including people in political leadership, with the technical understanding and know-how to take advantage of this technological revolution to improve the ability of our public institutions to deliver what people need,” Weinstein said. “And this kind of teaching is essential for that.”

Sarah Newman is collecting secrets. Specifically, she’s collecting secrets about AI. 

At conferences and other events, she puts out a wooden box and invites passersby to anonymously disclose the ways they’re using the powerful, mushrooming technology.

“I’ve written a speach [sic] with AI on a topic I don’t have a clue about,” wrote one anonymous confessor. 

“Using AI to write official policy language without necessarily disclosing,” wrote another. 

Newman, director of art and education at metaLAB at Harvard, sees a lot of uncertainty and angst out there. The aim of her project is to promote discussion about the personal use of AI at a time when its spread has already outstripped the ability of businesses and institutions to define what is appropriate in every instance.

“People don’t know what the rules are,” she said. “There are not established shared norms around both AI use — what’s appropriate and what’s not, across contexts — including when you should disclose use.” 

Since the 2022 release of the popular Large Language Model ChatGPT, businesses, schools, and universities as well as newspapers and publishing houses have been trying to sort it out.

But those efforts have been complicated by the technology’s defining property — its ability to mimic human output. This means it can be difficult, if not impossible, for institutions to police whether users are cheating. 

And there have been recent high-profile accusations of misuse.

In May, the British literary magazine Granta responded to allegations that one of its prize-winning short stories, Jamir Nazir’s “The Serpent in the Grove,” was AI-generated. (Nazir has denied using AI to write the piece.)

Separately, the author of a book about truth in the age of AI acknowledged using the technology in writing his book after The New York Times identified misattributed or fabricated quotes in the text.

In addition, news outlets and social media have produced a steady drip of stories about the use of AI in wedding vows, dating app messages, or eulogies.

The eventual societal fallout of all this uncertainty is unclear. But fields like journalism already have been affected.

“What we’re actually seeing is audiences assuming that real media is AI-generated, and that’s a really concerning trend to me,” said Andrew Deck, a reporter who covers AI in the industry for Harvard-affiliated Nieman Journalism Lab. “It indicates that [for] some audiences, in some ways, their default assumption these days is that something is false rather than the other way around.” 

In truth, Deck said, very little content at reputable news outlets is generated using AI. Reporters mainly use AI for things like transcribing meetings, scrounging up sources, or iterating title ideas. But they’re doing so amid historically low trust in news organizations.

Pew reported last year that about 56 percent of American adults had “a lot of” or “some” trust in information from national news outlets, down 20 percentage points from 2016. AI, with its reputation for spouting hallucinations with certitude, presents a challenge for an industry grounded in accuracy and pushing back against what Deck called a “trust crisis.” 

AI use labels, Deck said, “feel like a way to indicate to readers or audiences that maybe they should have an additional layer of skepticism.”

But disclosure might not always be a clean solution. “AI-assisted” could mean, in theory, anything from a quick source search on Perplexity to an editorial pass that was more like a rewrite.

Issues over uncertain standards appear to have even filtered down to families.

Leah Plunkett, Meyer Research Lecturer on Law at Harvard Law School and faculty associate at the Berkman Klein Center for Internet and Society, is an expert in digital technology, youth, and families.

In her work with K-12 schools, she’s noticed that kids are on high alert for signs that adults in their lives have used AI when they themselves have been forbidden it. 

“They have an antenna, as kids and teens — particularly teens — for potential hypocrisy or double standards,” she said. “It can be hard to get them to understand that actually different roles in a school or community ecosystem have always been permitted, and at times required, different tools and resources, because they do different activities.”

Newman of metaLAB has been fascinated by secrets, and why we feel the need to keep them, for a long time. A 2016-2021 project examining why we share so much with our digital devices, “The Future of Secrets,” was exhibited internationally. She moderated a metaLAB panel on AI, secrecy, and disclosure this spring. 

Newman believes that norms for AI use will become clearer in the coming months and years. She added, however, “We shouldn’t expect norms to just emerge. We should reflect on our own use and then help shape the norms we want to see.”

Like many people, Newman said she has a love-hate relationship with AI. But, she said, “People shouldn’t feel a need to lie. We should be curious about — even intrigued by — our own mixed feelings, and work toward normalizing appropriate, responsible AI use.”

Newman credits her former research assistant, Olivia Tai ’22, with the idea for the AI secrets project.

Tai is an artist and facilitator who has published her own collection of anonymous confessions. She hopes the project can encourage people to talk more openly about their use of AI and be more curious and understanding of how others view the issue.

“Like with all taboo topics, the more we shame each other, the less productive our conversations become,” she said. “We stop hiding when we feel some acceptance, understanding, and recognition from others.”

On May 29, the Office of Management and Budget (OMB), along with federal grantmaking agencies, proposed an overhaul to the government-wide framework that has governed for more than a decade how federal grants are reviewed, awarded, and managed.

The proposed rules to the Uniform Guidance, the framework’s formal name, would significantly change the way federally funded research is selected and overseen, giving political appointees a larger say in funding decisions and expanding the government’s ability to cancel grants already underway.

The proposal drew nearly 500,000 public comments before the window closed on July 13, with a large majority opposing the new rules. A final rule could come as early as Oct. 1.

Harvard submitted comment letters from the University, the Medical School, and the T.H. Chan School of Public Health, which urge OMB not to move forward with the proposed changes.

The Gazette spoke with John H. Shaw, senior vice provost for research, who wrote the University’s comment letter, about what the proposed changes would mean for researchers and for the federal research partnership. This interview has been edited for clarity and length.

---

What has historically been the purpose and scope of the Uniform Guidance? Can you give us an overview as to what’s happened recently?

The Uniform Guidance that’s issued by OMB is a framework that governs how the federal government allocates taxpayer dollars to fund its grant-making to research universities like Harvard, to other nonprofits, and to a wide range of organizations across the nation that the federal government seeks to support.

This framework represents the “rules of the road”: It ensures that those dollars are spent efficiently and effectively, and are consistent with the federal government’s expectations in providing that resource.

The scope and the breadth of the proposed changes, and the broad community they would impact has, I think, inspired such a strong response.

It’s typical when a proposed change to the Uniform Guidance is issued that the federal government calls for comment so that the various stakeholders have an opportunity to respond. The government, in turn, has a legal obligation to respond to those major criticisms — either to revise the policies or proposals or to explain why they’re not going to do so.

How do the proposed changes alter the current framework? What is the scope of these revisions?

They are vast in scope. They would affect more than a trillion dollars of federal spending across the range of areas in which the federal government awards grants and resources to help support the general public.

The document is hundreds of pages long. It’s very dense, and as a result, it reflects an amalgamation of a lot of different things that the administration has sought to accomplish through executive orders and other means over the last two years. 

There are areas that represent a fundamental change in terms of how research is conducted at universities, at affiliated hospitals, and other types of entities. That is ultimately what has motivated Harvard to respond in a way that is thoughtful and responsible. There are a couple of areas that really are the underpinnings of the ecosystem of innovation that the federal government has supported over the last 75 years in higher education. 

That innovation ecosystem is the reason that U.S. universities are world leaders in so many areas of technology and science more broadly. People feel as though these changes, if implemented, threaten to significantly reduce the competitiveness of the United States. They could have very direct and immediate impacts on areas of national importance — like national security, the health and well-being of patients, and the development of crucial technologies that we need to maintain our competitiveness and continue to support a vibrant economy.

How has Harvard responded to these proposed new rules?

Well, naturally, Harvard would have responded to something of this scale and this potential impact, but that was actually strengthened by a chorus of interest and concern that came from almost every part of the University community — from our faculty, researchers, students, and  alumni — recognizing that these changes would have dire consequences for the U.S. research enterprise.

We were certainly motivated to plan a formal comment, and we did that by reaching out to the people in the University ecosystem who would be impacted. We received a tremendous amount of very constructive and thoughtful input, which we used to inform the University’s formal comment to these proposed changes. 

In addition, the Medical School and the T.H. Chan School of Public Health have submitted their own comments focused on the specific concerns of their communities. Many other individuals, as researchers, as faculty, have also submitted responses. And finally, we’ve aligned with major national associations that support higher education and research, like the Association of American Universities and Council on Governmental Relations, that also recognize the threats posed by these recommended changes and want to have a thoughtful and aligned response.

You mentioned two key focuses in the University’s comment letter, the first being the introduction of new political oversight of federal grants. Can you tell me more about this?

This is probably the most impactful change. The relationship that the federal government has had with higher education and other institutions to conduct research has been successful because it placed the merit of proposed work as the standard by which decisions are made to allocate resources. 

It’s important to clarify that the federal government and the various agencies that help support this science always determine what the priorities for the nation are. There are program managers and others who work in these agencies who consult with stakeholders and determine that these are areas that warrant federal investment, and those areas are not static. They change over time.

In my 30 years as a federally supported researcher, in my own lab, the priorities for national research have varied substantially, and that’s good. That’s healthy. That’s something that needs to continue to evolve. 

But what’s being threatened here is that once those priorities are established by a funding agency — that we want to prioritize research on a particular form of cancer, or on a particular area of technology — the decisions about what awards to make would be political, not based on merit.

That will have dire consequences. Making something political means that those priorities are going to change depending on the nation’s political leadership. Research is not something that survives well when it has to be turned off and turned on with great frequency. It needs to be sustained. 

I’d also say that merit-based peer review is a process that is very challenging to get through.

Many federal agencies that award grants have a less than 10 or 15 percent success rate. So, really, to get an award approved based on merit is a very high bar. It means that your proposal has gone through scrutiny, often in a highly specialized area of research, by the world’s leading experts in that field — it is a very strong process. 

The answer to any individual investigator isn’t always favorable. Most of the time it’s not. But it motivates you to work toward an even higher standard. This is the best way to allocate taxpayer money to achieve the goals of the agencies and the federal government.

The other focus you mention is federal agencies’ expanded authority to suspend or terminate grants. Can you talk more about this?

Research doesn’t often hinge on a single eureka moment. It’s usually something that needs to be sustained over years to make important breakthroughs and to innovate. The ability, once awarded, to have confidence that you can conduct that research over the term of the award — whether that be three years or five years — is critical, because you need to bring on staff and students to help enable that research, and we, as an academic institution, make multiyear commitments to our researchers. 

And so, if there is this prospect that political changes or whims will turn research on and off midstream, it limits the University’s ability to invest in, and sustain, the most important part of the research enterprise, namely the students, staff, and faculty who enable it.  

Frankly, this uncertainty will make faculty risk-averse. They won’t be able to bring in as many students, bring in as many researchers to conduct as wide a range of research, because they’ll have this threat hanging over them that their grants can be terminated. 

This would not include grants that are terminated for cause. That capacity already exists and should be maintained. The OMB’s suggested shift would permit midstream terminations that really reflect changing political winds, which would truly threaten the opportunity to fulfill the potential of the research that’s been supported.

This is not only a waste of taxpayer dollars, but poses the risk of long-term harm to research outcomes. If you fund something and only get 25 percent along the way before it’s canceled, we never have the potential to realize the benefit of that investment — and affected patients and the public never get to benefit from that research.

What are indirect costs? And what does the proposed rule say about their role in funding?

To many, indirect costs are one of the least understood aspects of supporting a research enterprise. Essentially, these are funds within each federal research grant that support the costs of the research’s associated administrative and facilities expenses. We have seen, over the last few years, threats to reduce indirect cost recovery to very small percentages.

What’s proposed in the OMB guidance is something different because, essentially, those attempts to just reduce it to smaller levels have already been stopped by the courts and by Congress. The OMB guidance is suggesting that a university’s indirect cost recovery rate is something that can tip the scales of whether that award should be made to that institution or another. 

In contrast to direct costs, like salaries, equipment, materials, and supplies, which are assignable to research activities on a specific project, indirect costs are other essential expenses that reflect the true cost of that work — costs like building infrastructure, laboratory facilities, electricity and utilities, and administrative staff to support administration and compliance.

Indirect costs are critically important to the conduct of research, but they aren’t easily attributed to a single project, because they support state-of-the-art facilities and labs and the infrastructure necessary to ensure that research is conducted safely and in compliance with federal and state laws and institutional policies, ranging from environmental health and safety to research security.

The facilities are where the work is done. Without them, work could obviously not continue. And of course, the administrative costs are necessary because we need to conduct our research in ways that comply with the rules that the federal government has set for us. And we’re committed to adhering to those because they are critically important — like how we treat human subjects who are engaged in research. 

The critical challenge with using this rate as a thumb on the scale in deciding to award funding is that the administrative part of this is already capped so the differences between institutions are really in the facilities.

If you start awarding federal grants preferentially to institutions that have lower indirect costs, it means one of two things: It means you’re awarding them to places where the cost of buildings and infrastructure is less, or it means you’re awarding them to institutions that are less willing to invest in their own research enterprise. 

At Harvard, a lot of the research we conduct is lab-based. It requires major University infrastructure investments to have the capacity to do the research, and indirect cost recovery is a way that we partially recover some of the costs associated with making those investments. And so, if we disincentivize making those investments, universities will make fewer of them, and will have less capacity to do research, and certainly less capacity to innovate.

In addition, we want to ensure that decisions to fund proposals are based on scientific merit and transformative potential, not what costs the least.

What problems could the new rules present for scientists and researchers?

Well, for the scientists and researchers themselves, it comes back to this issue of sustaining their work — whether they are running a lab and looking to hire research staff, or whether they are students who are trying to meet academic degree requirements and have the capacity to go on and start their own labs in other universities or other research institutions across the country. It’s a commitment, of course, that we make to those students when we admit them — that we’re going to be there to support them through that process.

Moreover, the subjects of this research are important, deepening our knowledge of the world, while also creating new technologies, medicines, systems of understanding and other insights that lead to practical benefits for people in their everyday lives.

If grants can be terminated based on changing political considerations, that creates tremendous fiscal uncertainty, and so we may be forced to admit fewer students. We may hire fewer postdoctoral fellows.

The lack of certainty regarding the commitment that we make to these researchers will also undermine our ability to recruit the best talent from around the world, which is so critically important to the U.S. research enterprise.

If you look at any university, including Harvard, the impact of international students and researchers is profound. It’s to the real benefit of our nation that we maintain the opportunity to attract the best and brightest, because they help enable this innovation, which obviously benefits our nation and our allies across the world. The kinds of changes proposed by OMB would sow unpredictability, and therefore they would clearly reduce the scale of our research enterprise, and therefore its impact.

Are there examples you can share as to how these proposed changes could slow or halt current research, and how that translates into impact? How do rules that disproportionately affect universities like Harvard harm our national scientific competitiveness?

For the second question, research is a collective enterprise. We at Harvard are the beneficiaries of awards from many federal agencies, and we take very seriously the responsibility to conduct that research thoughtfully and capably.

We also, through many areas of research, interact with other research universities and institutions across the country, and in some cases around the world. And again, that’s truly important to enable the best and brightest minds to come together to solve some of the most important challenges that we have in terms of breaking through barriers in biomedical research and advancing critical technologies like quantum science. 

The ability to arbitrarily terminate an award in one place will not be isolated in its impact. It will have a ripple effect through this ecosystem. Threats to universities like Harvard will be felt by our important research partners across the nation, which exist in almost every state, and of course span organizations beyond universities, like research hospitals. 

To address the first part of the question, there are just so many examples where the outcome of research that’s had a major positive societal impact wasn’t clearly foreseen at the inception of that research. So much of our fundamental research takes decades to come to fruition, and examples of that illustrate the need for a sustained commitment to the best and most promising fundamental research. 

Its potential can’t be foreseen by a political appointee or someone who is not intimately familiar with the area of science and technology. But someone with deep scientific and technical experience can — through a merit-based review process — assess whether this research will help gain fundamental understanding of a system, biological or physical, that allows us to decipher the code to unlock doors that are closed to us right now.

That will have value, not only in generally increasing our knowledge, but potentially enabling us to translate that research in impactful ways for society.

What are a few examples of this research?

A major revolution in biomedical treatments today results from our ability to do gene editing. Technologies like CRISPR, of course, are examples of one of the fundamental ways in which gene editing is enabled.

Some of the most exciting and impactful research in this area today had its foundations, at least in part, at Harvard University, through faculty at Boston Children’s Hospital who were affiliated with Harvard Medical School. In one example, our scientists helped develop a capacity to turn on and off fetal hemoglobin in ways that could allow us to treat the blood diseases sickle cell anemia and thalassemia. 

It took decades from the time that fundamental research was conducted to the point where these therapies were available to patients. Now, of course, their success has been widely recognized and heralded. They’re not just incremental advances. They represent a fundamental change to the way we treat these diseases that are very, very impactful to patients — with their roots in fundamental research. 

In quantum sciences and engineering, fundamental technologies that enable us today to develop quantum computers and quantum sensors were a relatively niche area of physics decades ago. But they were supported by the federal government, and by research universities, for their potential to contribute to fundamental insights, and ultimately have proven to have such important societal impact in ways that we might not have originally foreseen.

One of these is in solving complex problems of molecular interactions that can help us unlock insights into optimizing the development of therapeutics to treat disease.

They also can help us, of course, in the context of securing streams of data. Quantum is widely understood to have the capacity to do code-breaking, but it also holds potential to help us transmit information in secure ways. It’s critical to the security of our nation, and this is why these types of programs are supported not only by the National Science Foundation, but by the military and other parts of the federal government that, like us, view them as vital areas of research. 

Twenty or 30 years ago, when we were laying the foundations of research projects like these, no one could have foreseen their full potential. If they were susceptible to political winds — if these projects started and stopped, started and stopped — we simply would not have reached the degree of impact that we currently have. And this is true in an area like quantum where, frankly, it’s come much faster than many of us anticipated it would. 

Another final example is work done at the Wyss Institute for Biologically Inspired Engineering that has developed technologies for something called an “organ on a chip.” These are physical devices that enable populations of cells to work together, akin to complex tissues within the body in performing specific functions, allowing scientists to test those cells’ response to different drugs or environmental factors. They’re valuable in many ways, including that they offer an alternative approach to animal models, one that’s more specific to human biology.

This is an area of research at the University recently funded by the federal government in which we explored how astronauts are exposed to radiation in space and what the impacts of that could be, and how that could be mitigated. And these devices flew on the recent Artemis mission.

As Harvard experienced the wave of mass federal grant terminations in 2025 — and we saw all the destructive effects — this project was an example of something so vital that you couldn’t just stop it. We had living cells, human cells from the Artemis astronauts, on these chips that were planned to launch with this mission. If we had not been able to sustain that work, we wouldn’t have been able to participate in that critically important mission for the nation.

What are the next steps in the comment process?

OMB is legally responsible for reviewing all the comments that have been submitted and to respond to them, either modifying their proposed rule changes to address those concerns or explaining why they’re not making the changes.

What’s remarkable here, again, is both the scale of the comments that they have received and the speed at which they want to implement these proposed changes. We think it’s critically important that the comments that we provided, that our Schools have provided, that our community members have provided, and those that have come from across the nation, are considered carefully and thoughtfully. It’s critically important to consider all feedback, and to recognize that it’s a consultative process.

And so, we’re certainly encouraging the time be taken to get this right. It’s important that we do that. It’s critical for the nation’s interests and to the many Americans who stand to benefit from the next treatment, cure, or advance in technology.

An academic superstar, her devoted husband, two kids, and dog pack up in the Bay Area of California and head out on a great American road trip to start the wife’s yearlong visiting professorship in rural Vermont.

So begins “Country People,” the fifth novel by award-winning author and psychiatrist Daniel Mason ’98. The story centers around the husband, Miles, as he tries to embrace small-town life and finish the dissertation he’s been nursing for 14 years, all while battling his own wandering sense of purpose and place in the world.

In an interview with the Gazette, Mason discussed how looking closely at some of life’s small but beautiful moments can be rich fodder for storytelling. Our conversation has been edited for clarity and length.

---

What was the inspiration for the novel? 

When I wrote my last book, “North Woods” (2023), I spent a year out in Western Mass., Southern Vermont, and upstate New York doing research, and there were a number of characters there that remained with me even after I was done.

I also found myself continuing to think about the comedy of our family of city people picking up for a year and trying out rural life — with all our preconceptions and fantasies about what that might be.

I started writing little scenes, not thinking it would actually become anything. Over time I was more drawn in by Miles and after a few months I decided, “OK, I’m going to try this.” 

The book is very playful. Was the creative process the same?

Definitely. My other books use humor from time to time, but it’s often accidental or incidental. “North Woods” had a bit more, but I’d never deliberately set out to write a comedy.

At first I was quite worried about that — writing something contemporary was already new for me, but a comedy? The idea of telling a joke and it falling flat is a horrible thing that we can all relate to, but 300 pages of jokes falling flat would be really embarrassing.

That held me back, but the more I wrote, I enjoyed it a lot. Miles turned out to be the kind of character who went anywhere, and it was a lot of fun to follow him — I’d be laughing to myself as I wrote.

The reader gets to witness the internal dialogues of many of the characters as they struggle to connect. How do the characters find — or fumble — relationships with one another? 

Miles is a person who is very open to people, but at the same time, there’s a loneliness there.

A theme that emerges early in the book is the difference between the connections he has in his life versus the connections that his wife has in her life. She has a very rich network of friends, from her academic circles, social networks, back home, or in town, whereas Miles has sort of gone his own way while still being very devoted to the family. 

I didn’t have to go very far to notice this pattern in my own life and a lot of my peers’. As I was writing the book, I kept seeing articles and academic papers about male loneliness and the struggle for men to find real connection.

Miles is interesting because he does find friends, but they’re like a bunch of little islands. They’ll connect over their trees, their gardens, or (later in the book) a shared interest in the Hollow Earth legend. But loneliness is pretty persistent.

“Miles turned out to be the kind of character who went anywhere, and it was a lot of fun to follow him — I’d be laughing to myself as I wrote.”

Throughout the book Miles is on an unending journey to finish his dissertation. In one sense he gets nothing done, but he also has a very full life. What does this tell us about him?

This is something that one sees in academia all the time: People are drawn earlier in their life to academia and are passionate about an idea, but that’s very different than fully executing a project.

This is something I’ve experienced, where I might be excited about something but maybe it’s not the thing I want to do with my life. That’s Miles. He’s a person who’s really interested in almost anything he comes across but still doesn’t know how to find his place. 

How did you find the right voice for this novel?

I think it developed over time. The voice is pretty different from my first, mostly historical novels. 

“North Woods” covers a wider span of time with different historical voices, and later in the book, in a chapter about an amateur historian, Morris Lakeman, I found myself writing in a contemporary voice, in what’s called free indirect discourse — which is third-person access to a character’s inner thoughts with the emotional inflection as they might have them.

I love this voice. “Madame Bovary” (one of my favorite books) is considered the great classic example; Virginia Woolf uses it magnificently as well. As I tried it out in “North Woods,” I loved it, and in this book, found it fit the wandering thoughts of Miles really well. 

In using that voice, the reader is privy to all of these unfiltered glimpses into what characters are thinking and feeling. What’s the benefit of that insight? 

Often those little anecdotes can say so much. As a writer, I’ve moved from detailed physical description to focusing more on little things that might shed light on a person’s personality. 

Earlier in my career, I had a prejudice I should write about “big” things, like war, love, nature (and more recently, climate change).

But in this book, I found that looking even at seemingly mundane moments could offer a different kind of insight. Like, for example, when Miles has this trivial question about whether he should wear jeans or sweatpants to meet a friend at his apple orchard for the first time. It’s trivial, but we learn a lot about him: about his indecision, his uncertain relationship between himself as a city person versus his fantasy of himself as a country person, about belonging, about insecurity.

That was one of the unexpected parts of writing, this opportunity to look at the stuff of everyday life — a kid’s school play, a trip to the emergency room — and see what light it can shed. 

One part of the novel involves a community of people who believe the Earth is hollow, and they spend their time looking into a local legend. What did you feel that added to the plot?

Many of us these days wonder: Why do people believe things that aren’t true? I didn’t set out thinking that the “Hollow Earthers” in the book would answer that question.

Initially, I wanted to find a legend that would draw Miles in. Later, there is another character, known as Snowflake Bentley, whom Miles meets in the emergency room, who voices some of the more theoretical questions about the taxonomy of imagination: Where does the boundary lie between phenomena that seem related? What is a conspiracy theory and what is a legend?

There is a continuum, and sometimes it’s hard (even in my professional life as a psychiatrist) to pinpoint the difference between imagination and delusion. 

How did you feel you grew as a writer, writing a comedy for the first time?

It was freeing. It gave me an awareness that the world around me has elements which, at first, seem simple — but only because I wasn’t looking more closely at them.

I worry a lot about the state of the world, and in the past have used history as a kind of lens to see it from a distance. Using humor provided a very different kind of lens. 

The SAT turned 100 this summer. The dreaded rite for high school students has changed several times since its debut in 1926, but Harvard was there from the start, trying to shape it into a reliable tool to gauge college applicants’ academic readiness.

The test is the best-known product of the College Board — a nonprofit organization formed in 1899 to standardize college admissions. Then-Harvard President Charles W. Eliot was a founding member.

The organization spent its first 2½ decades developing and administering exams on a range of subjects, but it wasn’t until June 23, 1926, that the first SAT — which at its inception stood for “Scholastic Aptitude Test” — was administered to more than 8,000 candidates. The test was elevated to its position as the standard college admissions exam in the 1940s. Today more than 2 million students annually take the SAT.

“The 100th birthday … it’s a question of whether you measure it from when the seed was planted to when it sprouted to when it became the giant tree that it is today,” said Andrew Ho, the Charles William Eliot Professor of Education at the Harvard Graduate School of Education. Ho is a psychometrician whose research aims to improve the design, use, and interpretation of test scores in educational policy and practice.

“In 1926 it was just a little test,” he said. “To me, as someone who does work in testing, it’s always about who uses the test for what purpose, and that’s really where James Bryant Conant and Henry Chauncey come in.”

Conant, Harvard’s president from the early 1930s through the 1950s, and assistant dean and scholarship director Chauncy were early boosters of the test. In 1934, Chauncey pitched the SAT to Conant as a way to find qualified scholarship students. Harvard began using the test for scholarship students in 1934 and, in 1941, began requiring it for all applicants.

“James Bryant Conant saw this opportunity to take it nationwide beyond just the boys in the prep schools — to look for talent throughout the United States,” Ho said.

The first SAT consisted of a 97-minute exam, featuring nine sub-tests on topics from math and reading to logic and analogies — made up of 315 mostly multiple-choice questions. And for much of its history, it continued in a similar format.

The most significant change to structure and scoring prior to the 21st century was the 1995 score “recentering” during which the College Board shifted the tabulation of raw scores to make them more reflective of student averages as the test grew in scale. This was also the first year the test included open-ended math questions instead of multiple choice in an effort to promote problem-solving over guesswork.

“Questions followed the curriculum, and less so this idea of aptitude as a set of games and mental puzzles divorced from what kids learned in the classroom,” Ho said.

The most significant pivots, though, according to Ho, were changes in the design and the marketing of the test as a college and career readiness test in the 2010s.

“The test became much less a series of games and puzzles and vocabulary words, and more a reflection of what kids learned in high school, that would prepare them ostensibly for college and later careers,” Ho said.

Over the last few decades, the SAT has become much more accommodating of open-ended responses, he said, and technology, including scannable text, has made it easier and more efficient to ask new kinds of questions such as short answers, or essays.

What hasn’t changed, though, is the public debate surrounding the issue of using a single test to measure a student’s academic capabilities. Critics have pointed to disparities in scores across race and class. As early as 1938, test preparation became a booming business for elites — undermining the College Board’s claims that the exam is “un-coachable” and an objective measure of scholastic aptitude.

Data gathered by the Harvard-based nonprofit Opportunity Insights found that children of the wealthiest 1 percent of Americans were 13 times likelier than the children of low-income families to score 1300 or higher on SAT or the alternative ACT tests, while the College Board’s publicly available data shows racial inequity.

In recent years, some schools became test-optional in their admissions practices. Harvard was test-optional throughout much of the COVID pandemic, reversing the decision in 2024. College administrators argued that the SAT is just one piece of the admissions puzzle, citing Opportunity Insights research that found, despite socioeconomic score disparities, the exam is a key tool for identifying promising students at less-resourced high schools.

“The data reveal that other measures — recommendation letters, extracurriculars, essays — are even more prone to such biases,” said Raj Chetty, the William A. Ackman Professor of Public Economics and director of Opportunity Insights. “Considering standardized test scores is likely to make the admissions process at Harvard more meritocratic while increasing socioeconomic diversity.”

Ho agrees with this assessment.

“For the vast majority of high school students in the United States, these tests already exist,” Ho said. “Those test scores already exist. Asking not to see them creates more inequality than it eliminates. Covering your eyes or asking others to figure out whether they should close our eyes for us, that creates this unnecessary dilemma.”

Ho calls the set of criteria that admissions officers look at — GPA, extracurriculars, essays, and recommendations, in addition to the SAT or ACT test — the “five-legged stool.”

“If you take a leg of the five-legged stool away, it rests on these four other legs, and we should look at those four other legs critically, think about their fairness, the equality of access to those educational opportunities to write good essays, to find good letters of recommendation, to be connected to good extracurricular activities, and to have good high school grades,” he said. “Those are also often restricted along lines of class and race and gender in ways that we have to be critical of and try to remedy as well.”

The most important thing to remember on the SAT’s 100th birthday is that the test is not the be-all and end-all of a student’s capabilities, Ho said.

“People who are at selective institutions should do more to demystify, clarify, and explain how test scores function in our system as one of multiple measures,” he said. “It’s a reflection of educational opportunity at a given point in time. That’s all.”

The state of the American social fabric — enormous, diverse, and rent by stark partisan divides — is daunting enough that it can be hard to think about, let alone try to mend.

At the Harvard Kennedy School, two social scientists are advancing the proposition that sports fandom, a powerful social phenomenon and force, may have a role to play in improving lives and strengthening societies.

That proposition is probably controversial. It’s certainly underexplored.

So a new project, launched by Professor Todd Rogers and doctoral student Audrey Feldman aims to fill the gap, with its first wave of findings set to publish in the months ahead.

Leaving aside the billion-dollar business, fandom is an undeniable social fact. In the U.S., nearly three-quarters of people say they follow a pro sports team, and the majority of those consider their fandom to be an important part of their identity.

(And on the week of the World Cup final, it’s noteworthy that 51 percent of people globally are fans of soccer alone. Americans are a bit less ardent (27 percent), but 40 percent planned to watch at least some of the games.)

Yet Rogers, Weatherhead professor of public policy at HKS, and Feldman have found that sports receive remarkably little scholarly attention. In top journals across the social sciences, only about 1 in 800 articles concerned sports.

The Fandom and Social Connection Initiative aims to change that.

Backed by a gift from FOX Sports and seeking partners among professional sports leagues, Rogers and Feldman have embraced the work as both an observational “green field” and as a target for social intervention — with a personal rooting interest for them both.

Feldman was drawn to HKS in the hope that behavioral science could improve human lives — for instance, by increasing access to underutilized social programs and promoting learning and social cohesion in refugee populations.

But amid conversations with Rogers on international development, Feldman said, “we realized we’re both big sports fans.”

Feldman, who completed a bachelor’s in economics at Notre Dame in 2024, didn’t escape the pull of Fighting Irish football, though her dad went to rival Michigan. And Rogers’ Mount Auburn Street office is dotted with Philadelphia Eagles gear.

“It’s social infrastructure in a world where there are few — if any — institutions left that are this wide, and deep, and diverse.”

Todd Rogers

The initiative attempts to apply rigorous modern tools to their shared passion.

Implicitly, the work involves a reclamation project for sports fandom itself, which Rogers argued is not only understudied but maligned.

“We did a survey where we asked non-fans to identify fans on that ‘ascent of man’ scale,” Rogers said. Presented with that famous succession of increasingly hominid silhouettes ambling to the right, most non-fans placed passionate sports fans somewhere around Homo erectus.

To Rogers, that’s a dangerously low appraisal of fandom’s social role. “It’s social infrastructure in a world where there are few — if any — institutions left that are this wide, and deep, and diverse,” he said.

With many forms of social organization — social clubs, political parties, and churches — in a kind of secular decline, sports fandoms loom larger than ever in their power to combine and motivate large groups of people, across racial, economic, and political divides.

Consider Congress. Nineteen of the 50 current state delegations are made up entirely of one party or the other.

By contrast, Rogers and Feldman found, the reddest NFL fanbase (the Tennessee Titans) is still politically diverse, with two registered Republicans for every Democrat, while the bluest — the Chargers of Los Angeles — is roughly the inverse.

And — as anyone knows who’s ever watched a game in a packed stadium or bar — shared fandom can serve as an instant de-polarizer.

In a paper currently under review, Rogers and Feldman applied game theory to test the effects.

In a computer survey, they presented participants with hypothetical “counter-partisans” — people who hold views on politics and culture starkly opposed to their own — and asked them to award a “bonus” to one or the other.

If they knew one “counter-partisan” also rooted for their favorite team, that one got the “bonus” at rates that outstripped those for counter-partisans with a shared race or ethnicity, education, or home region.

Only a shared religion came close to being that influential, which seems revealing.

Often enough, sports fandom is as inherited as any other marker of identity. But — like religion — it’s also both active and elective: a seasonal, ritual allegiance that people can either outgrow or choose to renew throughout their lives, Rogers argued.

And often enough, they do the latter — even when their favorite team is luckless, perennially “rebuilding,” or to all appearances cursed.

To Rogers, that opting-in suggests that sports fandom isn’t motivated principally by on-field events but by the bonds that form around them with parents, siblings, colleagues, and neighbors. (“How else would there be Cleveland Browns fans?” he joked.)

Sports fandom — like religion — is both active and elective: a seasonal, ritual allegiance that people can either outgrow or choose to renew throughout their lives.

The positive social effects of being a sports fan have not totally escaped serious notice before this summer. Rogers credits the book “Fans Have More Friends,” co-written by Ben Valenta, head of strategy and analytics for FOX Sports. And he speaks monthly with Daniel Wann, professor at Kentucky’s Murray State University, whom Rogers calls the under-appreciated “godfather of psychology and fandom.”

Wann gave him a useful metaphor. “He said shared fandom is like a shared photo album,” Rogers recalled. To follow the same team as another person — even a perfect stranger — is “to know where [they] were in February 2018, and where they’re going to be next September,” he said. “Our hearts are going to be synchronized.”

(Literally, Rogers added: Some psycho-physiological results suggest that spectators’ EEGs start to synchronize around the highs and lows that punctuate big games.)

Still near the outset of their shared work, Rogers and Feldman are still theorizing why sports are such an effective social solvent — say, in conversations with a stranger. But it makes sense. Relative to the weather, they’re engrossing, human, and almost inexhaustible. And compared to politics or current events, sports are safe.

That’s usually true even in cases of acrimonious rivalry, they argued. Feldman spoke of the fun of watching a football game even with fans of Ohio State. “I think their team is the worst, but at least we can talk about football,” she said. “There’s a mutual respect there.”

Rogers noted one draft finding that backs that up: in which fans of rival teams in the English Premier League said they’d sooner socialize across the aisle separating, for example, Man United and Liverpool than with people who don’t watch soccer at all.

Even as their theoretical work is ongoing, Rogers and Feldman are already pursuing opportunities to intervene in the real world.

Their first published work together was an op-ed published last fall for TIME, arguing that the football-averse should embrace a team in America’s most popular sport. NFL fandom, they write, is a “simple and universally accessible medicine” in an age of loneliness.

That work will continue this summer, in partnership with the Boston Red Sox’s AAA affiliate, looking for ways to nudge spectators into conversation with their neighbors in the bleachers.

Though their papers are still being written up, the first indications support the idea that sports can serve as a kind of social glue, Rogers said.

“Audrey and I have experiments where we can induce 40 percent of participants to watch a game with others, and — at the end — they feel more connected, and in a way that persists for weeks,” he added.

Just as this country celebrated its 250th birthday, some geneticists were celebrating America, too.

At the end of June, the National Institutes of Health announced its “All of Us” Research Program had become the world’s largest database of integrated health and genomic information. The project, launched in 2015 under the Obama administration, aimed to collect comprehensive genetic information and health records from at least a million Americans. Eleven years later, they’re approaching that goal, with 747,000 volunteers sharing their whole genomes, lifelong health records, or both.

Comparable nonprofit efforts had already taken shape in the U.K., Canada, Germany, and Japan. But Alicia Martin of the Broad Institute hopes the U.S. repository will be the most valuable resource of its kind for the study of disease’s genetic and environmental roots, given the nation’s size, diversity, and patchwork healthcare system.

In this edited conversation with the Gazette, Martin explains the value of such a resource in a country that’s home to over 342 million genomes and counting.

---

I know you didn’t work on this project directly but what excites you about “All of Us”?

No, I wasn’t directly affiliated with the project. I would love to give some kudos to some colleagues — like Stacey Gabriel, Niall Lennon, Heidi Rehm, and the team they led at the Broad Institute — who’ve done an enormous amount of work to generate this data, including the majority of the whole-genome sequencing.

I’m a population and statistical geneticist, so I work on large-scale biobanks around the world to understand the genetic, phenotypic, and social determinants of health that impact disease risk overall. So mostly I’m excited as a researcher who’s using the “All of Us” database.

It’s been around a quarter-century since scientists — including Eric Lander, founding director of the Broad Institute — first “mapped” the human genome. How does a database like “All of Us” build on that work?

We’re way beyond where we were with the Human Genome Project. There, for $3 billion, we sequenced one genome — that’s three billion base pairs of DNA. Here, we have the full genomes of over half a million individuals, deeply sequenced.

And this is an incredibly comprehensive resource, so it also includes electronic health records that are longitudinal, allowing us to understand a disease’s course and trajectories, rather than simply who has a disease now. And it includes multi-omic data.

What does it mean for the bank to contain multi-omic data?

Well, in high school biology, you might have learned the central dogma that DNA makes RNA makes protein. Your DNA is fixed from birth.

But other parts of the “multi-ome” are dynamic. They change throughout your life course, and they differ depending on your cell types, what age you are — all sorts of things. So your multi-ome is a much more comprehensive capture of what’s happening in your body at any given time.

“A resource like ‘All of Us’ will be beneficial for treating everybody, even if a given discovery was made using a subpopulation in the bank.”

There are, by now, many biobanks operating around the world — the U.K.’s is famous, for example. What is different about this one?

It’s true that global biobanks are abounding, and each one is really a national treasure, given how much they can teach us about disease biology. You mentioned the UK Biobank, and that has an amazing amount of whole-genome sequencing, whole-exome, other multi-omic data as well, in addition to an incredible depth of electronic health records going back decades.

But “All of Us” has a unique focus on drawing on representative populations from the U.S. And the unique thing about the U.S. is that our ancestors come from all over the world. With that, we’re able to identify novel biology that we wouldn’t necessarily see if we were only studying populations in the U.K., say.

One of the poster children of the genomic drug development pipelines is PCSK9 inhibitors, which lower LDL cholesterol levels. And they’re a proof of the power of representative genetic study. They came out of discoveries for loss-of-function variants that were identified at much higher frequencies in African American populations, in a few thousand individuals in the Dallas Heart Study. It took far, far larger numbers of individuals of European ancestry to validate that loss-of-function finding in large biobanks elsewhere.

So this is not necessarily just about knowing how to treat a specific subpopulation for specific diseases. In fact, a resource like “All of Us” will be beneficial for treating everybody, even if a given discovery was made using a subpopulation in the bank.

In many cases, the genetic mutations that increase risk of a disease are not sure things — the mutated gene may interact with other exposures or conditions so that the disease takes root, or it may not.

Can a biobank heighten our level of granular understanding, to know about all that a given participant’s life that might have caused or prevented health problems?

I think resources like this are so valuable because they’ll push really hard on what we’d call the “exposomics” space: In other words, all of the things you’ve been exposed to in your life that, as you said, can combine with genetics to affect your total disease risk.

For example, when we study proteomics — the proteins your body produces based on genetic instructions — we can develop a quantitative, objective measure of, say, how much you’ve smoked over your life course, and how the cumulative biological load of that affects your overall disease risk. Same with alcohol use, or pollution exposure through PM2.5 measures.

And in the end, you can integrate all of that information to try to understand better. And conversely, sometimes people can have very protective lifestyles or habits that mitigate their underlying genetic predispositions toward disease.

We’re interested not just in whether someone gets a disease, but whether and how that disease progresses, or whether they respond to therapeutics. And all of those things are to some extent informed by the new types of data that “All of Us” has released.

This would seem to have a new usefulness as gene therapies begin to become more commonplace.

Yes. We have these incredible examples of biomedical successes. Like Victoria Gray — an amazingly brave woman, a sickle-cell patient — who got the first CRISPR gene-editing therapy. She underwent clinical trials with CRISPR editing in 2019, and that paved the way for Casgevy [a gene therapy used to treat sickle-cell disease and beta thalassemia, both inherited blood disorders].

Or Baby KJ, who had an incredibly rare mutation — one in more than a million babies — that is lethal in about half of infants. He had personalized CRISPR editing therapy that effectively cured him. He got to graduate from the NICU.

But those involved millions of dollars of therapy for treatment of a single individual, and treatment that might only work in a subset of patients. “All of Us” can help us quickly identify that subset of patients for a given disease, and we can turn it into a learning platform that teaches clinicians which patients they should be on the lookout for, to target these therapeutics to the patients who need the treatment and who will respond.

You could see how a national biobank would be much easier to build in a country with a national health service of some sort: standardized forms, standard procedures, just one bureaucracy to deal with. How did this get done in the U.S., with its private healthcare providers, private and religious hospitals, not to mention the rural and military and Indigenous contexts?

The team of researchers behind “All of Us” took electronic records as they were. And you’re right — it’s inherently incredibly messy in this country, which has a highly fragmented approach to healthcare. We have to take everyone at their word.

But wherever people are getting care, that generates diagnostic and billing codes. And you can use billing codes to try to understand genetic predispositions to all sorts of diseases.

For example, you know who has Type 2 diabetes, since that’s been diagnosed in their health record previously. And I want to applaud participants, who were — I think — exercising a high degree of beneficence by allowing us to see their entire health records, sharing a lot of personal information. We owe them.

Lauren Groff stunned her interviewer at a recent Harvard Radcliffe Institute talk when she revealed she has published poems under a pen name.

“I do write poetry and I publish it, but not under my name,” said the acclaimed author of short stories and novels with mischievous laughter, in response to a question by fellow writer Gish Jen ’77, a 2002 Radcliffe fellow, who was moderating the first installment of the Virtual Radcliffe Book Talks series.

“Whoa,” said Jen. “Oh my goodness. This has not come out before. An alias … fascinating.”

Groff, a three-time National Book Award finalist, is the author of the novels “Monsters of Templeton,” “Arcadia,” “Matrix,” “Fates and Furies,” and “The Vaster Wilds,” as well as the collections of short stories “Delicate Edible Birds” and “Florida.”

The 2018-2019 Suzanne Young Murray Fellow at Radcliffe has in the past credited poet Emily Dickinson as the reason she became a writer. At last week’s talk she said that she reads poetry in the morning to start the day with “a jolt of literary caffeine.” But she has said very little of her own poetry until now. Groff didn’t disclose her pseudonym, and Jen didn’t ask Groff to reveal it.

“Poets can do things that I think prose has a really hard time doing,” said Groff. “In poetry, the speaker can float; it doesn’t have to be an obvious speaker. There’s a lot of compression that has to happen in a poem that sometimes you can’t really do it in a novel or short story.”

Among Groff’s favorite poets are Terrance Hayes, Natalie Diaz, Ada Limon, Walt Whitman, and of course Dickinson, but she said she is fond of way too many poets to list them all. She praised Devon Walker-Figueroa’s 2025 poetry collection, “Lazarus Species.”

The conversation also touched on Groff’s writing process, inspirations, and “Brawler: Stories,” her latest collection of short stories that deal with themes of love, fear, and violence. The stories in “Brawler” were inspired by Groff’s exploration of the “overall macrocosm of violence in North American cultures, particularly the United States culture, and the way that it can be reflected often in the family unit.”

Of her writing process, Groff said she likes to work on different projects at the same time. “If I obsess too much about one thing, that thing will never get done,” said Groff, who co-owns with her husband The Lynx Books in Gainesville, Florida, where she lives. “But if I can sort of partition out my obsessions into three or four different projects, then I feel as though I’m not putting all the weight into that one thing.”

Groff has worked on some books for decades, not only because some stories take time to develop in her subconscious, but also because she sees them as “living entities” that need time to grow and take shape, like children.

“One has to allow them to grow and to come to you with the amount of time it takes for them to grow,” she said. “You can’t force a book to do what it needs to do. You can’t force it to come to you in its wholeness.”

Groff said she has learned to wait. “Maybe it’s not the right time to tell it, maybe you haven’t found the right form in which to tell it, or maybe you’re just not wise enough to tell it yet,” she said.

A graduate of Amherst College, Groff earned an M.F.A. from the University of Wisconsin-Madison. Groff spoke fondly of her time at Radcliffe, where she came up with the idea for “Matrix,” a historical novel based on the life of medieval writer Marie de France.

“Radcliffe radically changed my life because it gave me that book and sort of the triptych of books that I’ve been working on since 2019,” said Groff.

Published in 2021, “Matrix” is the first installment of Groff’s triptych followed by “The Vaster Wilds,” set in Virginia in the early 1600s, which came out in 2023.

Asked about the book that could potentially complete her triptych, Groff shared she is planning to explore religion, a topic she finds both fascinating and moving because “each religion in the world shows humans struggling and grappling with things that we cannot possibly understand.”

“What I wanted to do with this is to wrestle with the ideas of Western religion and the way that religion and the way we treat the natural world is like the way that we treat women,” said Groff.

“I started in the medieval time with Marie de France, and then I went on to a Protestant coming to the New World, and this very hungry capitalist way. And the third one is set now, and that’s all I can say about it. But it’s still wrestling with God. I’ll never stop wrestling with God.”