Cory Doctorow wants you to know it doesn’t have to be this way.

At a discussion Wednesday night moderated by Elizabeth Mynatt, dean of Northeastern’s Khoury College of Computer Sciences, the Canadian-British writer and activist addressed a friendly room at the Berkman Klein Center.

Doctorow has been a longtime comrade in the fight for an open, safe, and humanistic internet for which the center was founded 30 years ago.

But he also belongs to another, older lineage: one of enraged populism and antitrust militancy.

At the event, Doctorow issued a call to arms against the digital age’s new monopolies. And he cast doubt on the possibility that AI, the tech industry’s latest “disruption,” would improve human lives or justify its costs as — to date — “the money-losingest thing the human race has ever done.”

He also warned that the much-touted potential of AI to change the world, for good or for ill, may miss real risks the technology poses.

Recently Doctorow achieved a one-word breakthrough with “enshittification,” a neologism describing a three-step process by which big tech platforms lure in users with promises, say, to respect the privacy of their data, betray them to business clients — then betray those business clients, too, in service of their own bottom lines.

The end state, Doctorow argued, may be the worst of all possible worlds: for users, services they’ve grown to hate using but struggle to leave, for companies, a “brittle” equilibrium, with the ever-present threat of a mass exodus.

It’s a partisan story, but then Doctorow is a partisan — he’s been an activist affiliated with the Electronic Frontier Foundation for more than 20 years. It’s also a plausible one, as user dissatisfaction grows with a series of once generally favored tools — from Amazon Marketplace to Google Search — even as tech stocks keep on rising.

In any event, “enshittification” has been a hit. The American Dialect Society named it “Word of the Year” in 2023, and it has since been added to the Oxford English Dictionary (“major bucket-list stuff right there,” Doctorow said).

That isn’t in spite of the word’s unprintability in polite outlets, Doctorow argued, but because of it: “There’s just something about the minor license to vulgarity — especially when you’re very angry … that seems to unlock a willingness to listen to a more thoroughgoing policy analysis.”

And woven in between schoolyard jokes and righteous indignation, Doctorow offers that analysis: a political economy of Silicon Valley firms incentivized not to do one thing well but to sell the public on Next Big Things.

A “mature” public company, he notes, grows slowly, pays regular dividends, and supports stock prices of 10 to 20 times their annual earnings per share.

By contrast, tech companies — the archetypal “growth” businesses — sustain price-to-earnings ratios twice that high, often with the promise that they are about to conquer or absorb some existing profitable sector: Think taxis or trucking, bookstores or newspapers.

And, when much of their top talent is at least partially paid in company stock, it becomes vitally important that those promises are believed.

The latest promise is the subject of Doctorow’s latest book, “The Reverse Centaur’s Guide to Life After AI,” published in June. He referred to it as “the greatest literary blunder of his life,” as his attempt to wave away most AI panic instead invited more angsty conversation.

At 55, Doctorow is arguably one of the world’s oldest and most venerable “digital natives”: a godfather of blogging and open-source software developer, co-founder of Boing Boing, and a Wired contributor.

And he’s had moments of awe at the “super-cool, interesting” things the latest AI models can do: from generating their own stochastic collages to finding and transcribing a memorable quote from a recent podcast.

These are “centaur” experiences, where a technology endows a human user with superhuman strength, expediting their work and improving their lives. (It’s a tale as old as corrective lenses, Mynatt pointed out.)

But Doctorow worries that — with AI — many humans will have reverse-centaur experiences, where employers, aiming to lower labor costs, force them to use the technology on far-from-human terms.

For example, he speculated, a hospital might lay off all but a few of its human radiologists, with the remainder “marking the AI’s homework.”

Given the algorithms’ speed and the statistical plausibility of their findings, those few remaining humans will be looking for errors that are devilishly hard to spot — and will take the blame whenever mistakes are made.

Asked about the potential existential risks that trouble AI engineers, Doctorow scoffed. He argues that LLMs, at bottom “word-guessing machines,” still cannot research themselves, remain relentlessly error-prone, and are “nowhere near recursive self-improvement.”

His own AI worry is the firms’ growing centrality in the unequal global economy.

Doctorow noted that the seven companies making big bets on AI now represent 35 percent of the total value of the S&P 500. That value could be “vaporized overnight, when investors lose confidence,” he said. From there, he argued, the story is all too human: financial crisis, austerity, and a further lurch into hard-right politics.

Doctorow’s first writing as a teenager was science fiction. He still publishes in the genre, and in “Reverse Centaurs,” he celebrates it as an “anti-inevitabilist” literature.

In sci-fi, “many different writers have imagined many different futures, each of which can feel possible and even desirable,” he writes. And “the fact that social relations between people and their technology can be different means that the current arrangement is a choice, and —­ crucially —­ it means that we can choose something else.”

If digital life is less fun than ever — characterized by worsening user experience, false promises, ubiquitous advertising, and threats of doom — “it’s not because of the iron laws of history,” he said. “It’s because named individuals in living memory decided this was the environment we should have … But we can make different decisions.”

The participants at a John F. Kennedy Jr. Forum debate last week on “diversity” and “pluralism” in higher education were clearly suitably matched.

Kevin Roberts is president of the conservative Heritage Foundation think tank. His group published Project 2025, a 900-page policy document that has heavily influenced President Trump’s second-term agenda and pushed for the end of diversity, equity, and inclusion (DEI) programs.

Cornell William Brooks, a civil rights attorney, formerly led the National Association for the Advancement of Colored People and is currently Hauser Professor of the Practice of Nonprofit Organizations and professor of the practice of public leadership and social justice at Harvard Kennedy School.

The two did agree on this one broad principle: Universities should be composed of people from a wide range of backgrounds.

But Brooks called the concept “diversity”; Roberts preferred the term “pluralism.”

And the pair sharply diverged on how far the country has progressed on racial discrimination and whether race-conscious programs are still needed — a divide that likely came as no surprise to Brooks, who said in his opening remarks, “When the school asked me to debate, I asked them to ask Kevin Roberts.”

Asked by moderator Laura Meckler, a Washington Post reporter, what higher education has gotten right and wrong in their efforts, the two outlined these disagreements.

Roberts said race-conscious programs had outlived their usefulness.

“In 1964 or ’65, did we need programs — that then weren’t called DEI — in order to level the playing field? Perhaps so,” he said. “But our problem at Heritage, my problem as a scholar in this work, is that we are denying the progress that’s been made over the intervening several decades.”

Brooks said, “When we talk about racism, antisemitism, xenophobia, Islamophobia in American society, we’re not talking about some distant age in the past. No new administration will bring about some new era in America where we don’t have to be vigilant with respect to fighting bigotry of all forms.”

Meckler pushed both speakers to defend their arguments.

She asked Roberts to rationalize how the current administration could argue for a smaller governmental role in education while also aggressively “telling universities and K-12 schools what to do.”

Roberts said the administration is trying to correct injustices and argued that if Harvard doesn’t want to listen to the government, the University should reject federal funding, which he did as president of Wyoming Catholic College.

Meckler asked Brooks to respond to the idea that some students won’t be admitted because they don’t check the right demographic box.

Brooks said that many white students at Harvard benefit from kinds of advantages not available to most Black and brown students. He also argued that conservative voices are overly focused on scholarship programs for Black and brown students, ignoring financial aid earmarked exclusively for members of other racial, ethnic, or religious groups.

Summarizing the event, Roberts said the disagreements between the two speakers came down to an “honest-to-goodness policy disagreement about the kinds of policies that work.”

He argued that race-neutral policies that emphasized socioeconomics and geography would better serve the public.

Brooks invoked the story of civil rights attorney Pauli Murray, who was rejected by Harvard Law School because of her gender but ended up crafting the legal strategy that cleared a path for the landmark Brown v. Board of Education ruling.

To live in a racially diverse democracy, Brooks said, “We have to fight for it every day, all the time.”

Catherine Snow is the John H. and Elisabeth A. Hobbs Research Professor of Cognition and Education at the Harvard Graduate School of Education.

Perhaps cursive went out of style because there was a sense that it was just a technical skill, and one that was becoming less and less necessary with the universal availability of keyboards. There’s this sense that there’s a “problem” of putting stuff on paper, and the technical solution is that you type it and it looks neat and it’s well spelled and we’re not wasting time doing something that kids always hated — namely, practicing cursive. I could see that as an argument that could be convincing.

The problem is, in the process you are losing something that’s really of value: important cognitive challenges.

It’s quite a challenge for very young children to understand how letters are formed and how they are sequenced in words. If you’re not really concentrating, you might end up spelling “cat” C T, for example, because those are the salient sounds in cat. If you type words wrong, you’ll get spellchecked and you might never realize that you were wrong.

Take a slightly more complicated word — let’s say “moderate.” Well, there’s no way of inherently knowing what’s in the middle of moderate, right? It could be M O D A R A T E, or it could be M O D I R A T E, and if it just gets spellchecked, then you never actually come to make the connection. Having to spell it yourself, letter by letter as you write the word, draws your attention to the appropriate sequence of letters.

There are a number of studies which show that there are more areas of neural involvement in handwriting than in typing. There’s more visual cortex involvement, as well as all the differentiated motor skills that go into forming letters. I think it’s really the increased attention to the sequence of symbols that generates automaticity, and automaticity is a key element in good reading.

But this advantage of facilitating learning to read from handwriting comes both with cursive and with printing. It doesn’t really matter. I’ve never encountered anything that would suggest that proper cursive with all the letters connected is better than printing.

I had the proper eight years of elementary school Palmer Method training: The first 15 minutes of every school day were copying sentences 28 times on lined paper. But now I never read cursive, and I find that I really have to work hard if I’m in an archive or a museum and I’m trying to read letters exchanged between people in 1945. It’s a skill that kind of declines without practice. That’s one of the big advantages of learning to write cursive: In the process of learning to write it, you’re learning to read it. It’s learning to access history and literature and shared culture.

Should we keep cursive? I think we should. I’m not deeply dedicated to cursive, but I think we should spend more time in schools having kids write. There are kids who never learn cursive who can function with a pen in their hands perfectly well. They should be writing with pencil or pen on paper, and we should be focusing much more on the content of what they write than correcting errors. You could learn cursive later, if you really wanted to. There is not a critical period after which you can’t.

— As told to Sarah Lamodi, Harvard Correspondent

Assina Kahamba and Maralmaa Chinzorig both arrived at Harvard for graduate school with the goal of getting the necessary training to help their communities with big problems.

For Kahamba, it was strengthening teacher education in Democratic Republic of the Congo. For Chinzorig, it was improving medical care in rural Mongolia.

Both received grants from the Harvard Innovation Labs (i-Lab) Navab Social Impact Fellowship Fund to help apply what they learned to urgent problems at home.

Assina Kahamba

Kahamba didn’t plan to become principal of an all-girls’ school.

After finishing a computer engineering degree at the University of Kinshasa, she went to work in the private sector in the Democratic Republic of the Congo, her home country.

Eventually she began thinking about the big problems plaguing public education in DRC and decided she wanted to shift her career and focus on making it better.

Kahamba was working in an elementary school, responsible for the children’s character development, when her school’s principal abruptly resigned. Suddenly, she found herself in charge of all the teachers and students.

One of the biggest issues Kahamba found involved incomplete teacher preparation. Some had solid knowledge of a subject but poor pedagogical skills. Others were skilled at breaking down and explaining new information, but had critical knowledge gaps.

“I was in charge of providing a good education to the students,” Kahamba said, “and I was seeing my teacher unable to give a class on fractions.”

She looked for professional development programs nearby but couldn’t find any. She started one in her own school but discovered she didn’t have the background to make it effective. Ultimately she decided that if she wanted to train teachers properly, she needed more education herself.

To get it, she looked to Harvard Graduate School of Education, where — with funding help from her community — she began a one-year master’s degree in global, international, and comparative education.

Her professors encouraged her to rethink her preconceptions of teacher training.

Fernando M. Reimers, the Ford Foundation Professor of the Practice of International Education, told her old-school programs that bring educators together for just a couple of days of training have become obsolete. For better results, teachers need to be coached over time, with ongoing evaluations of effectiveness.

Kahamba’s studies took her beyond education. Through entrepreneurship classes, she learned how to run — and sustain — a business for training teachers in the DRC.

“With this context of cuts of aid to Africa, it was very important to know how to have money as a business,” she said, “and not only expect money coming from the U.S. and Europe.”

With funding help from the i-Lab’s Social Impact Fellowship Fund, which provides about $200,000 in support for a dozen ventures started by students across the University, Kahamba was able to assemble a small team to help build out her vision with Koyekola Academy, the initiative she founded.

When she returned to the DRC after graduation, some people were surprised. “Everybody was saying that this girl had gone to the U.S. — she won’t come back,” Kahamba said.

But educators from across the country, including some who had helped fund her education at Harvard, wanted to find out what she had learned. She was surprised when two directors came from Kisangani — more than 700 miles from Kinshasa — for the training she started to develop at Harvard.

A few months after graduation, Koyekola Academy has three contracts with schools — each of which will receive ongoing training over months of work. Her goal is to have a successful pilot runand work with the government to expand the program across the country.

“We want to build a community of teachers who really want to change the world,” Kahamba said.

Maralmaa Chinzorig

Growing up in rural northern Mongolia, Chinzorig heard harrowing stories from friends and family about trying to get medical care.

The nearest provincial hospital could take two or three days to reach during the winter. When dirt roads were impassable, residents crossed the frozen Khuvsgul Lake for better treatment.

They then arrived at local hospitals where doctors had only sparse paper records and no access to digital ones. Without knowing details about their patients’ backgrounds, doctors struggled to provide good diagnoses.

Often, frustrated patients would take any generic Soviet-era drug recommended to them by peers, even if it didn’t address the problems they were facing.

“It’s almost like an ancient, tribal healthcare system,” Chinzorig said. “It’s really shocking for people who live in urban places.”

After graduating from college in Ulaanbaatar, Chinzorig worked with the Ministry of Health in Mongolia, where she helped integrate technology into urban hospitals.

After seven years, though, she realized that she would need better project management skills if she wanted to push reforms through the government bureaucracy. She decided to attend Washington University in St. Louis for a degree in business administration.

Chinzorig was struck by the level of well-being Americans expected to have into old age. Where she grew up, she said, people expected to suffer from a long list of diseases above the age of 50.

She became determined to create a company that would ultimately challenge the government to do better.

But while Chinzorig had ideas for the digital innovations she could try to implement in hospitals, she thought she could use more training to learn how to work more effectively with the governments to help expand healthcare services.

She decided to enroll in a master of public policy program at the Harvard Kennedy School. In courses about negotiation, Chinzorig learned skills that would help her push for improved healthcare when she returned to her home country.

“I think my first year of Harvard made me become someone who is a huge advocate and brave enough to challenge the system,” she said. “You guys are telling me the system is just, but what is the standard of just?”

With funding and support from the i-Lab’s Social Impact Fellowship Fund, Chinzorig has expanded the offerings of her company, MeTools, a digital-health platform that allows clinics and hospitals to share medical records without reliable internet.

The company has helped the National Center for Zoonotic Diseases — which tracks infections like the bubonic plague — sync the records of 13 rural facilities across Mongolia.

At Harvard, she said she’s learned how to bring together different stakeholders, listen to their pain points and interests, and arrive at solutions to which everyone can agree. In Mongolia, and across the broader region, she hopes to bring these skills to use.

“Someone who knows the technology is stuck on the technology. Someone who knows disease surveillance is focused on disease,” she said. “I’m trying to become the leader who can connect and listen to all kinds of different people and have this collective solution.”

Excerpted from “The Art of Fighting: The Transformative Power of Conflict” by Priya Parker, M.P.P. ’12, published on Sept. 8 by Riverhead Books, an imprint of Penguin Publishing Group, a division of Penguin Random House LLC. Parker is a specialist in conflict resolution who works with organizations, communities, and leaders.

When people in conflict first call me, they tend to say some version of: “Make this go away.”

I get it. Conflict is not fun. They are concerned, worried. A group of people they care about, or rely on, or live among, is in crisis.

Make this go away.

What I have learned from years of fielding these calls is that most people, including people very savvy in other areas of life, have an unhelpful mental model of what group conflict is. Perhaps they think of it as a court case, and they’re calling me, their lawyer. Make this go away. Or they think of their fight as a story about to break in the media, and they’re calling me, their crisis communications whiz. Make this go away. Perhaps they think their conflict is just troublemaking by some members or employees or parishioners, and they’re calling me, a heavy, a tough guy who is going to find the complainers and ask ominously if they’ve got a problem. Make this go away.

I try not to judge this impulse. I get it. But I also inform my callers: I don’t make things go away.

I have a different perspective. To me, “How can I make this go away?” is the wrong question to ask, because trying to stave off a fight is not always what best serves a group. Thinking of conflict as something to bury or put back in the bottle or disappear is common among group amateurs. The artful fighter knows better. Some tears in group life should indeed be quickly sewn up and forgotten about. But others are the beginning of a big, revealing, and necessary reckoning, of a struggle that will force a group to face fundamental, half-​buried questions about who they are, what they are for, and how they operate. These conflicts are unlikely to go away and will keep coming up again and again if left unaddressed.

So when people ask me to make a fight go away, I tell them it’s time to stop being afraid and start becoming curious. Why is this happening? Why is this happening to us? Why is this happening now? Why am I feeling so bad?

Rebekah called me when she realized her family Passover ritual was broken. Her extended family had been gathering for the seder for some 30 years. Her grandmother, a Holocaust survivor and the family matriarch, always insisted that the family stick together. They were, as Rebekah put it, “desperate to cling to each other” and Passover Seder was the “big central thing for the whole family.” The ritual is a more-​than-​2,000-​year-​old tradition of remembering and telling the story of the Jews leaving Egypt and finding their freedom. For Rebekah’s family, “there was a lot of symbolism about slavery and redemption that for the Holocaust survivors was a really important way to express their grief,” she explained. It was often the only moment when family stories of survival were brought up at all. When she was young, Rebekah cherished the ritual and the gathering. It shaped her and so much of her family deeply.

Year after year, the whole family would witness the youngest among them ask the famous four questions of the seder. It was where family members shared important news — like a cousin who one year came out at the table. (During one hymn, the family would always joke and rhyme and make up new passages. That year, during the refrain “El bneh, el bneh,” Rebekah’s cousin Alex blurted out, “I am gay!” and without missing a beat, another family member retorted, “El bneh, el bneh, and that’s OK!”) It was where all the cousins would introduce their partners to the rest of the clan. When a niece’s chair broke under her and her new fiancé didn’t instinctively rush to catch her, the family knew the relationship was over. It was where Rebekah herself announced a pregnancy, one that she later lost.

But Passover was also where she started feeling the suffocating pressure and judgment around having kids. It was where she got stuck in the role of cooking for weeks ahead of time and throughout the festivities. And then once she finally had kids, she started to notice how sick they would all become after the late nights and frenzied days of family time. She started to feel the consequences of taking off days of work to prepare for the holiday, and then more precious days off to stay home with sick kids after the fact. It was becoming unsustainable. Like at many family holiday tables, ideological differences were also becoming more apparent. An uncle would regularly hit “reply all” to the family welcome email thread and send “long political right-wing rants,” which he would then bring to the dinner table. Then COVID‑19 hit, and the family couldn’t gather at all.

Until that moment on our Zoom call, Rebekah had been very matter‑of‑fact — almost clinical — in describing the situation. And then she got quiet. I saw that she was softly crying. When she heard the clacking of my typing stop, she looked up. “I’m so sorry,” she said. I said gently what I say on many calls: “Please don’t apologize.” And she just cried. “I don’t want to feel like this. I wish I didn’t care so much,” she said. She was articulating an unspoken assumption shared by so many people on these calls: If only she could just sidestep this part of the process. If she wasn’t bogged down by feeling so sad, she could act!

“Feelings aren’t just OK here,” I told her. “They’re crucial. They’re not an inconvenience. They’re information. They’re a natural resource. They’re energy.” As Rebekah was rationally plotting her exit from her family’s celebrations, she viscerally remembered how deeply she cared for these people.

Feelings, no matter how painful, are not bad. They are grist. You may not be able to solve what’s in front of you if you can’t be bothered to cry. Tears are the lubricant of group conflict resolution. “No More Tears” may be a great formula for kids’ shampoo, but it’s a disastrous approach to adult fights.

Rebekah was experiencing anticipatory grief trying to figure out what she should do with this messy group that she loved but where the formation was no longer working for her (and, she suspected, for others as well). In a study conducted for the National Institutes of Health, researchers found that our brains process turmoil in relationships and perceived isolation with the same pain receptors as physical attack. Often, people don’t know what to do with these feelings.

Your feelings about a group conflict are valid and normal. But they’re more than that. They’re useful and informative. In fact, when a group feels really stuck, feelings are a signal of possibility.

When you find yourself feeling fragile or hurt in the aftermath of some kind of rupture, it may be that you are experiencing something that the conflict resolution specialist Sethu Nair calls “relational longing.” It is one of the most important superpowers you can tap into when seeking to resolve and transform a conflict.

Nair is a veteran group dialogue facilitator who works at the Center for Creative Conflict Resolution in New York City. She’s the kind of facilitator who gets a call when a “flag war” breaks out in the offices of a local government agency — Black Lives Matter signs on some desks, American flags or Blue Lives Matter ones on others. Would she be willing to come in and help the staff confront the divisions within?

Nair is not intimidated by garden-​variety conflict. She’s a founding member of Hidden Water, an organization that developed trauma-​informed circles for individuals and families impacted by child sexual abuse who wanted to heal with others.

In explaining how she begins to approach a conflict, Nair says that she must find the “relational longing” at the center of the fight. This relational longing, she says, is “the only avenue of possibility” through the fight, and she believes it exists everywhere. Relational longing is the residual desire within us to connect to someone or something, even when we’re hurt or mad. “We wouldn’t even have conflict if we weren’t trying so hard to be heard,” Nair told me. “I love finding that, and it’s not so hard to find if you really listen for what people are longing for.”

We don’t fight within our groups because we hate each other and want nothing to do with each other anymore. We fight because we are yearning to be in relationship with each other still, perhaps just not in the form of relationship we are currently in. (When that’s no longer true, we don’t fight; we leave.)

Where there are strong feelings, there is energy, and energy is easier to work with than apathy. There’s a there there. Where there is some beating pulse of relationship left, even a badly arrhythmic one, there remains a possibility of transforming the bond.

The inverse is also true. If a rupture stirs no feelings in you, that is useful to know. You may have a problem in the group, but no longing to solve it.

By the end of the calls I get, if I’ve done my job well, the person feels better. Nothing may have changed in the situation, but they have begun to learn what the artful fighter knows: that when faced with the first flushes of a conflict, they must steel themselves and refuse to sweep back what is peeking out from under the rug. They must push themselves to be curious — about why they’re feeling what they are, about what areas of heat are causing their group to fracture, about what they may have failed to notice before, and about why all this may be happening now.

Once the laments recede and the real questions arise, the artful fighter is well on their way. What awaits them now is trying to figure out, through all this questioning, what they have on their hands.

Copyright © 2026 by Priya Parker

What do we want from AI?

That was the question posed by Alex Pascal, executive director of the Berkman Klein Center for Internet & Society, at the outset of a panel discussion co-hosted by the center on how AI “can serve people and make our society better.”

The powerful, potentially game-changing technology has been in the headlines in recent months after high-profile reports of experimental AI agents going rogue and hacking into other systems. Industry leaders have warned that unchecked AI development poses dangers to humanity.

AI’s role is inherently double-edged, said Jeff Dunn, chief operating officer of the Digital Trust Council, which co-sponsored the event.

“We want to cure diseases but in order to do that we need to use a tool that can create bioweapons,” he said. “You can’t just put a knife through the middle and hope for the best.”

Amit Goldenberg, associate professor at Harvard Business School and a member of its AI Institute, broke down the kinds of artificial intelligence now in use, and the range of the roles the technology can fill.

At its most basic, he said, “it’s a tool, like autocorrect, and it doesn’t have a personality. The other extreme is a full-fledged agent that has a name, that has a personality that you can interact with.”

The “million-dollar question,” he said, is distinguishing where the different AI tools and models fall on the continuum and which ones are best suited for specific needs.

At the heart of this issue are two questions: “What are the type of interactions that will improve humans’ well-being and what are the type of interactions that will harm them?” Goldenberg said. “I work under the assumption that humans need humans.”

Avijit Ghosh, lead technical AI policy researcher at Hugging Face, said that we need more accurate means of evaluating AI systems in order to develop a greater sense of public trust in the technology. Hugging Face, a platform and library for AI developers, was hacked this summer by Open AI agents that went rogue. Ghosh said he couldn’t comment on the cyberattack on his company.

He noted that we are in an era of “benchmark maxing,” or posting more supposed benchmarks than necessary in order to boost public trust.

Dunn chimed in, saying that evaluations need to be done properly. “If they’re not validated or not audited, then it’s just a marketing stunt,” he said.

As a remedy, Ghosh recommended a system that measures both system behavior and its impact on people.

“It’s also not an unsolvable problem,” said Ghosh, who works on what current evaluation methods can and cannot capture. “We have decades of research in human-computer interactions.”  

How precisely to conduct such evaluations was open for discussion.

Ghosh, for example, warned against anointing certain panels or experts to judge the value (or harm) of any AI agent.

Most regulations and evaluations currently come “from the top down,” he said. “I want the general public to have access because we don’t know what we don’t know.

“If we collectively, as a society, decide that these are the 20 organizations that are allowed to evaluate models and nobody else [can], then we are probably depriving ourselves of really interesting insights. Something interesting can come out of a bug report if the public were allowed to submit some.

“We need better ways for the general public to participate in the general reporting and assessment.”

AI companies, he said, could be incentivized to respond to such feedback. “Maybe in exchange, the government can say, ‘You will not have liability if you fix this reported problem within 60 days.’”

“If we can align the business with what humans would want, then I think we will be OK,” Dunn said.

By now David Malan may be one of Harvard’s most recognizable (human) faces.

Since 2007, Malan — a Gordon McKay Professor of the Practice of Computer Science — has built CS50, “Introduction to Computer Science,” into the University’s largest course.

During that time, the online version of CS50 has taken on a global scope. Over the years, millions of students have registered in CS50x or its growing menu of more focused modules: on cybersecurity and law, Python, SQL, and AI.

Malan is on a sabbatical this fall, his first since taking over the class, with teaching duties passing to Henry H. Leitner. In his absence, the teaching team will continue to draw on the course’s latest improvement: the CS50 Duck, an AI tutor that aims to guide budding computer scientists to an answer without doing too much of the work itself.

Malan and the CS50 team have documented the Duck’s progress in articles detailing the bot’s 2023 rollout and their ongoing efforts to train it to become a better teacher.

In this edited conversation, Malan tells the Gazette about those efforts, the upsides and downsides of AI learning, and the value of computer science in an age of “vibe coding.”

Gazette: It’s been a few years since CS50 first introduced an AI tutor in courses, and you and your team have been tracking how it’s worked, the good and the bad. How did you decide to approach the problem?

David Malan: At first, it was very experimental. ChatGPT came out in November of 2022, and starting in the spring of 2023, we launched what we call the “CS50 Duck,” which is essentially an AI tutor.

We tried to leverage the ChatGPT API, in effect, to make a less-helpful version of ChatGPT. The Duck is much less inclined to just give students answers outright or generate code for them, and it behaves more like a good teacher or tutor: asking more questions than it answers and trying to guide students towards solutions.

How do you set up the Duck to play that “less-helpful” role?

It involves a couple of techniques. One is just what’s called “prompting” these days, whereby we just write literally English instructions telling the underlying ChatGPT agent how to behave. For instance, writing out statements like “do not give students outright answers to problems,” or “adhere to the course’s policy on academic honesty.”

But that alone tends not to be sufficient with large language models, or LLMs, like ChatGPT. So we also have a lot of code that we ourselves have written that tries to evaluate a response before the student sees it — and sometimes rejects or retries it, in order to get a more constructive answer.

As someone who has taken CS50 courses online, I can say that they are welcoming but difficult. It’s like a class in a foreign language; you can feel your brain straining to take on a new shape. But there’s evidence that the use of LLMs in education can smooth away the productive difficulties — the ones that help us learn.

We’ve definitely had some issues there — hence the iteration over the past couple of years. First of all, we were seeing “long tails”: really high usage among some students, where they might have asked something like 200 questions of the Duck. We don’t claim to know precisely where the line is, but 200 struck all of us intuitively as too many. No human teacher would invite a sequence of 200 questions before sending a student back to their desk to work.

So we added one mechanism that draws inspiration from games like The Legend of Zelda, with a “heart system” that decrements [as you’re injured in the game]. Within the Duck AI, we added a “heart system” that only allowed them to ask a few questions per unit of time, and that’s a knob we’re still turning.

At the same time, if there is an industry that has embraced LLMs by now, it’s software engineering; a recent survey found that almost 90 percent of software developers use AI in some form (though they have learned not to trust their output entirely). In short, this is the state of the art and there’s a lot of upside here, too.

Oh, absolutely. We wouldn’t still be using AI if we didn’t feel it was a net positive, educationally. It allows students to ask far more questions than they might previously have been able to, logistically or socially — and that’s very healthy. In my own time as a student, graduate and undergraduate, I remember walking into a professor’s office hours, and walking out no more enlightened, because I felt like I was overstaying my welcome or didn’t want to let on that I felt like a dummy.

It’s been a net positive for us teachers, too. There have been times in CS50 — and I imagine really any large intro course at peer institutions — where office hours had become unmanageable. In the early days, the average wait time could be as long as one hour, which was just not acceptable.

As a technologist, the flurry of news of the past few weeks — LLMs “solving” intractable math problems, or hacking websites unseen — did you see this all coming?

Instinctively, yes: It felt inevitable. But I didn’t think it was going to happen this quickly. It certainly seems reasonable to me that, as time goes on, computers will be programming themselves more and more. That’s been a trend for years. Back in my day, we’d make our web pages by hand, for instance: writing out HTML tags and later CSS. But then tools like Adobe Dreamweaver came along — applications that could generate that same sort of site without writing HTML, just clicking buttons. Then web-based alternatives like Squarespace or Wix.

So we’ve been eliminating the boring parts of coding for some time, and it seems inevitable that as our own programming capabilities get better, we will be able to program computers to write logical code. It’s sometimes imperfect but will get better and better. Therein lies some of the concern of late, and the danger.

The anxiety-provoking watchword is “recursive self-improvement,” which I think is intuitive even to the layperson. Once the most sophisticated LLMs can teach human students effectively, they can also teach themselves — perhaps to a point beyond human understanding or control. That prospect has some people quitting big AI firms and/or pushing for a pause on further development — how worried are you?

Well, I’m not sure it benefits any of us to just be worried without speaking up or taking action. I do think convincing humans to consider and to pass regulatory legislation along certain lines is probably wise — and sooner rather than later.

What concerns me most is not only that things like the Hugging Face incident can happen without our initially knowing about or understanding them — but how quickly they can happen. The computers we use and the code we write operates far faster, cognitively, than us humans. To me, that’s the dangerous part: because by the time even the most intelligent humans figure out what’s going on, it might be too late to fix something.

You’ve seen AI misalignment firsthand in CS50, as you detailed in that second paper. The Duck was told explicitly not to respond to students’ queries with blocks of actual code — and yet it still does up to a quarter of the time. What’s driving that behavior?

It’s a couple of things. Sometimes it’s just built-in randomness, which — as we teach in CS50 — can be a really powerful ingredient in algorithms. You don’t necessarily want a computer to find a solution to a problem and then always spit out the exact same solution, which might be correct but not optimal. So the Duck might occasionally be trying out a new response, randomly.

At other times, it’s just a consequence of the data that the model was trained on. There’s both good and bad input out there, and the model might have just “learned” to spit out code because that’s what people do on technical forums like Stack Overflow or on Reddit 90 percent of the time.

That has been one of our challenges, too, and it’s much better now. But LLMs aren’t deterministic in the sense that they will always do exactly as you command, because the reality is they don’t really understand you at all. They are just simulating understanding, based on their training data.

I think that’s one clear advantage to computer-science coursework: to understand how this or any new technology — as it changes our lives — actually works. Do you see any misconceptions out there you want to correct?

Certainly I want people to be skeptical of what AI tells them. That’s hard when, most of the time, these things do seem to be giving you the answers that you want, and it’s very easy to assign a personality to them.

But when you do internalize exactly what these technologies actually are, that hopefully snaps you out of that misapprehension, and you realize that you should approach the output of any computer — and even of many humans — with some skepticism.

I’d worry that, if people are not taught this, certainly in these early days of LLM-based AI, they might be led astray by misinformation that’s being spat out by these things.

For many years, the advice to young people was “learn to code,” and for a while it had some real credibility: high salaries, job security. But when someone armed with an LLM can “vibe code” an application in an afternoon, it looks a bit different. Do you think that’s still good advice?

The short answer is yes. Even before AI, in courses like CS50, we said consistently that the course’s purpose is not to teach students a given programming language, but to teach students to think more methodically, more algorithmically, to help them clean up their thought processes, to help them take in input and produce correct output. The programming languages that we use in CS50 have always just been a means to that end.

Ultimately, we want you to be able to look at these apparently magical devices and decompose them in your mind into their constituent building blocks, so you can better understand, and perhaps even be less fearful of, what they’re doing. That hasn’t changed.

Recommendation algorithms promise social media users a personalized experience, but some experts worry they promote doomscrolling, brain rot, and myriad mental health problems. Australia recently proposed legislation to allow users to opt out of algorithms with the goal of protecting children online and giving people more control over what they see on their feeds.

We asked faculty members from different fields of study to share their views and concerns about recommendation algorithms. Their responses have been edited for clarity and length.

It is standard to define manipulation as a form of trickery or covert influence. Some definitions suggest that manipulation treats people as “tools or fools.” My preferred definition sees manipulation as an influence that fails to respect people’s capacity for reflective and deliberative choice. The basic idea is that manipulators exploit a lack of information on the part of their victims or take advantage of our behavioral biases (such as a focus on the short term, unrealistic optimism, and limited attention). You can think of manipulation as a form of “dark nudging.” The problem is the same, from the Garden of Eden to the era of algorithms and AI.

I like recommendation algorithms! They can help us put our time and our money in the right places. If an algorithm tells me about a new book on behavioral economics or constitutional law, I will be grateful. Personalization is often highly beneficial. I don’t want to see anything about hockey, advanced physics, Jell-O, Beethoven, or Bach, even though a lot of people do. The concern here points to manipulation, as when recommendation algorithms exploit people’s lack of information or their behavioral biases. That can be a real problem.

We need laws and regulations to protect against the worst forms of manipulation, just as we need them to protect against deception. We need a new right — the right not to be manipulated. As individuals, we can and should pause a few beats and ask: “On reflection, do I really want to buy what they’re selling? With my time or my money? They might be trying to trick me!”

I get to do a lot of deep listening to young people, and usually our aim is not just to understand what they see when they’re behind their screens, but how they see it. I’ve been doing this work for more than 15 years. Recently, the “algorithm” seems to have shifted from a supporting player to a leading role in teens’ telling of their experiences with tech. They say things like: “My algorithm is just not great on that app” or “Lately, my algorithm is trash.” 

Young people shape their feeds in ways they’re well aware of, as well as in ways that they say they don’t understand. Their feeds shape them, too. One 16-year-old recently told me that her feed is filled with posts where, “It almost feels like everyone’s pushing the rush of your life — like get married at 18, have kids by 20. That’s just not everybody’s reality, but it makes it seem like it’s supposed to be.” Other teens have feeds that look completely different. 

When you sit alongside teens as they scroll, it’s abundantly clear how they are being profiled and pulled. Their interests (and their identities) are mirrored back with distortions like fun house mirrors. Another high schooler who is an avid gamer showed me his feed was full of “edits” (rapid-cut montages of movie scenes and TV shows), games, and ads for sports betting and online gambling. 

Algorithms can contribute to teens (and all of us) feeling inundated by narrowly themed content. For one 17-year-old I spoke with in the spring, the algorithm was saturating her feed in prom content: “I get it, I get it, it’s prom season, but it’s like every single time I refresh, it’s prom. But I don’t think it really changes how I feel about prom itself. It’s just like when I’m looking at it on my phone, I’m like, bro, can I see something else?”

Our personalized algorithms transport us into very different worlds, and even when we understand how they work, it doesn’t inoculate us from their impacts. For researchers, algorithms can also make it hard to capture evidence of the effects of these technologies, because everyone’s experiences are so varied and different by design. 

Responsibility for digital agency can’t rest just on the shoulders of our young people, who are being asked to withstand designs that prey on psychological vulnerabilities and developmental sensitivities. We need more control over algorithms, and features that allow us to make changes to them. Young people also deserve to understand how and why agency can be hard to exercise — in particular the incentive structures that are behind design choices that impact them.

Hyper-persuasion is a word I use to describe the compelling and even coercive parts of persuasion seen in the operations of modern digital media. An early example that serves as a harbinger is the 2014 Facebook Experiment by the Proceedings of the National Academy of Sciences journal, in which 689,003 users of the site found themselves unwittingly “nudged” into slightly more depressive (or slightly less despondent) states, as measured by their behavior after having their personal feeds manipulated. This was the power of social media — to induce “massive-scale emotional contagion,” almost like turning a dial on the user’s psyche. Little by little, as we engage with these technologies, we are ungrounded. The algorithmically-driven version of this is seen in how almost all of us — be honest! — can be prone to getting caught in a loop, a flow state, a doomscroll, or another daily type of dissociative mind state.

Social media interactions and AI chatbots, with their targeted algorithms, can become microenvironments. One of the key principles, really the foundation, of classical brainwashing is what Robert Jay Lifton called milieu control. Everything else — the emotional engineering of a captive, their ideological realignment (mostly temporary), or a radical personality change — follows from this control of the local environment, especially via control of what comes in and what goes out. Milieu, after all, means place or surroundings. You can imagine a highly controlling abusive cult or a prisoner of war camp that has no fences or barbed wire because it is so remote (this was the milieu in which U.S. GIs captured in the Korean War found themselves, where they underwent communist re-education). Although these circumstances seem extreme compared with the microenvironments created by your seemingly tiny interactions with your phone, for example, the same dynamics are at play. The device, with its algorithmic targeting, selects or controls what comes in and out of view for you, to which you react, which then continues to feed the selection process. This is why the word siloing is often used. It amounts to a type of invisible control.

My recommendation is to guard your attention very carefully. If it goes astray, or you find yourself in a mindless attention-sapping loop, bring your attention back to awareness. Strive to check in with your own feedback, as experienced in the body. There’s a subtle feeling that can arise when any of us encounters draining or (even) undermining messages. You can ask intermittently, how does this make me feel? It’s not a matter of avoiding what makes you uncomfortable but rather of making a deliberate choice that giving my attention to this is meaningful. 

An algorithm is any procedure that takes an input, performs some clearly specified computation, and produces an output. The first algorithms that we learn, in grade school, are how to add, multiply, and divide numbers. In machine learning, the process of training a model on data, either to predict (e.g., “Is this individual eligible for a loan?” or “Is this chest X-ray that of a healthy patient?”) or to generate (e.g., images, videos, text), is an algorithm called “gradient descent,” whose roots go back to 19th-century mathematics. Its input is data, and its output is a bunch of numbers, in some cases trillions of numbers, specifying a model. Somewhat confusingly, these models themselves are also algorithms. They are designed to take in some input, for instance a picture of someone’s face, or some prompt in plain English, and output a response, for instance whether the person has checked into their flight, or an answer to a math question.

Algorithms and scientific progress have gone hand in hand since antiquity. More recently, algorithms for numerical simulation have been instrumental in the study of complex physical and biological systems. Algorithms for search and sequence alignment enabled the sequencing of the human genome. Algorithms for reconstructing an image from noisy measurements are run every time a doctor performs an MRI or CT scan.

Recommendation algorithms are pervasive in social media. Their basic functionality is to take as input a user’s browsing patterns and output suggestions for content meant to maximize user engagement and drive sales. The implications for addiction, among other mental health concerns, are well documented. In terms of privacy, not only is it unclear to the user which aspects of their interactions with social media are being used to drive recommendations (“Is my phone listening in on my conversations?”), but these interactions ultimately also go into training the machine learning model that is making recommendations to other users, possibly revealing to them traces of a user’s activity they never intended to share. There is now over two decades of scholarship on this latter kind of vulnerability, and a massive body of work, e.g., on differential privacy, proposing interventions that algorithm designers could employ to mitigate these risks.

My overarching concerns about the impact of recommendation algorithms on young people’s lives are about two types of decisions: those that are made about young people without their knowledge or full consent, and those that are made by young people without their full awareness or meaningful consent, about the role of a recommendation algorithm.

Recommendation algorithms appear in tools, services, and platforms that might inform access to major life opportunities such as employment, insurance, etc., but also in social and emotional learning programs used by schools. And in social media, the feed being pushed is usually through the secret sauce of a recommendation algorithm, and for kids and teens, they might feel that a lot of the suggested content is organic and natural. The recommendation algorithm really can lure you into a false sense that this content is really meant for you. It can also get very insidious, even risky and dangerous, if the content that is being fed to you to keep your attention is depicting activities, emotions, experiences, relationship dynamics, and worldviews that are harmful or risky to minors.

I think the law in the United States has largely failed to provide ethical, practical protections for kids on social media and many other tech platforms. Ideally, the federal legislature would have done comprehensive individual consumer privacy for the digital age a couple of decades ago, and at the very least should have done it for minors, and they haven’t. We have seen many leading tech companies move fast and break things, fail early and often, act all the way up against, sometimes even past, the boundaries of what the law permits. So, to the extent that we have seen tech companies avail themselves of the freedoms entitled to them in this country to build lawful businesses, I would say shame on companies that have chosen to conduct their affairs past what the law permits, or even close to the boundary, and it is long past time for the places where that has happened to change.

I am not anti-tech. I am not anti-social media. I am not anti-innovation. I think the formal and informal governance mechanisms here are complicated, and I think that most people, whether they are federal lawmakers, state regulators, tech executives, or vice principals, are trying their best to regulate the way these companies operate and protect our kids and teens. I also think we could, and should, do much better.

Early this year, in remarks at the World Economic Forum in Davos, Switzerland, Prime Minister Mark Carney of Canada declared that there had been a “rupture” in the rules-based global system of military, economic, and diplomatic alliances by the West, an arrangement that has helped preserve peace for more than 80 years.

Carney’s speech arrived in response to recent major shifts in national security and trade policies by the U.S., long viewed as a linchpin of the system. The talk drew a standing ovation and widespread support from international political and corporate leaders at the forum.

In order to prepare future leaders amid rapid, widespread changes, the Harvard Law School recently launched the National Security Law & Policy Initiative. An expert in foreign relations and national security law, Kristen Eichensehr, Balthrop Cassidy Professor of Law, will serve as the initiative’s faculty director.

In this edited conversation with the Gazette, Eichensehr explains the initiative’s goals and the need to think about what comes next.  

Gazette: Why has the Law School decided to launch this initiative now?

Kristen Eichensehr: We are at an inflection point in national security. This is a time when things feel very unsettled domestically and internationally, and those two are not separate. What the United States chooses to do as part of its domestic policy has tremendous consequences internationally for individuals, for countries, and for the international order.

Harvard Law School has longstanding strengths in national security law, both at the faculty and the student levels.

On the faculty level, Harvard professors have been involved for decades in constructing the international order and working on national security policy in the United States. Among them are Manley Hudson, who was a U.S. delegate to the Peace Conference after World War I and became a judge in the Permanent Court of International Justice; Abram Chayes, who shaped the law during the Cold War; and Philip Heymann, who worked on the need to balance security and liberty in response to terrorist threats.

At the student level, we have incredible student organizations on national security, a leading national security law journal, and a national security law association. But what was really missing from that was an intellectual home for national security at the faculty level.

What are the initiative’s goals?

We’re hoping to use Harvard’s convening power to bring people together at a moment where national security is in a tremendous period of flux, both domestically and internationally. Part of the initiative’s goal is to marry together the domestic law piece and the international side.

The first year of the initiative is going to focus on the domestic aspect, including separation of powers and the role of Congress, in particular in national security policymaking.

The second year is going to tackle this question: Where is the international order headed? Canadian Prime Minister Mark Carney said that we are at a point of rupture in the international order, and I think that’s right. The question is: After the rupture, what comes next?

What are the most pressing national security issues on which the initiative will focus?

Since the 1970s, we have seen a tremendous expansion in executive power in presidential administrations of both parties, and we’ve reached a point where people sort of assume Congress isn’t going to push back against the president. It’s important to think about ways that Congress could reassert its prerogatives in national security.

We have seen moments before where the presidency and the executive branch expand power in ways that do finally prompt Congress to push back. The mid-1970s was a pivotal time for that. In the wake of the Nixon administration and the Vietnam War, Congress enacted the War Powers Resolution, the International Emergency Economic Powers Act, and the National Emergencies Act.

Part of what we want to work on over the course of the next year is: Are we headed toward a moment where that might be possible and desirable? And then more concretely, what does that actually look like?

For the purposes of the initiative, we’re going to convene a conference next spring to bring together academics, policymakers, and students to think through some of those questions, and we’re also going to convene a seminar that will bring students directly into conversation with thought leaders on these questions.

You said that the initiative will also tackle issues related to the post-World War II international order. How so?

I’ve been thinking about this quite a lot, in connection with the 25th anniversary of 9/11. It’s striking to me to think how different the world is now.

The only time NATO has invoked its Article Five, its collective self-defense provision, was to support the United States in the wake of 9/11. We’re now living in a world where NATO is at risk because the United States is putting the alliance into question, and where the U.S. president has threatened to seize territory from a NATO ally, in the case of Greenland and Denmark.

The world surely looks very different now.

In addition, the paradigmatic prohibitions in the international order are increasingly under challenge, including from so-called great powers such as the United States, which has been one of the greatest champions of post-World War II international order.

This is not a world where the United States is a totally dominant global hegemon. Even if the U.S. wanted to impose its views on the world, it’s not at all clear that it could, or that other countries would go along with that.

The scope of countries with which the United States will get its way or that it could persuade is narrowing. We are now in a trade war with Canada, one of our traditionally closest allies.

I don’t think the world is going to snap back to before 2016 or before 2024. We’re going to be somewhere fundamentally different, and we need to think carefully about what the U.S. national and international security policies should be.

New technologies and artificial intelligence are on many people’s minds when it comes to national security issues. How will the initiative address this subject?

Concern about and competition for new technologies, especially AI, are driving many national security decisions today. The U.S.-China tech competition is perhaps the biggest example.

I see these issues as deeply intertwined with questions about the future of the international legal order. Where interdependence and globalization used to be seen as stabilizing forces internationally, they are now viewed through the lens of security risk.

Tech competition is prompting the turn away from some international institutions, like the World Trade Organization, and toward nascent efforts to build new alliances among smaller groups to address issues like access to rare earths and critical minerals. 

Finally, what do you hope this initiative will achieve?

A fundamental goal that I have for the initiative is to think about how we can best educate the next generation of national security leaders and hopefully draw on their ideas even before they become national security leaders.

We are at a moment when innovative ideas and new thinking, some of which can come from students, would be helpful to the country and to the world. I’m hoping we can foster the school’s educational mission, help shape the next generation of leaders, and draw on the Law School’s convening power to bring people together and have some of these difficult conversations.

We’ve recruited an advisory board that is bipartisan. The board members have served in every presidential administration of the past several decades. I’m hoping we can bring people together in a way that will help to build some bipartisan support for particular ideas and hopefully some consensus to tackle the most pressing issues in national security law and policy.

Though it’s best known as an online retail and streaming giant, Amazon has also played a central role in the AI revolution.

In August, the company’s valuation hit $3 trillion. Much of that was driven by Amazon Web Services (AWS), the cloud computing operation that serves as a primary platform for both large-scale frontier model AI builders like Anthropic and OpenAI and for smaller firms developing simpler, more cost-effective AI tools for business applications.

During a talk Thursday at Harvard Business School, Amazon CEO Andy Jassy offered his view of where AI is going and rejected the idea of a winner-takes-all race from which one model emerges to set industry standards and control the marketplace.

“There is not one tool to rule the world — it never works out that way,” Jassy told Andy Wu, the Arjan and Minoo Melwani Family Associate Professor of Business Administration. “Customers always want choice, and in fact, most of the sophisticated generative AI applications today use multiple models.”

Ultimately, there will be six to eight AI models that compete with one another, Jassy said. Some may be a little better in one area than another, but they’ll all be useful and offer the level of intelligence, cost, access control, and safety customers want, he predicted.

One major obstacle AI still faces is the high cost of inference, the process of training models to make decisions and predictions. Though the number has come down significantly in the last few years, it’s “still too high,” Jassy said, largely because of the price of computer processing chips, a problem Amazon is seeking to address with its own line of budget AI-accelerator chips.

Jassy began working at Amazon three days after graduating from Harvard Business School in 1997. He was part of the team that founded and launched AWS in 2002. Those first few years were “by far the hardest, most difficult years of my career,” he recalled.

“None of us were sure it was going to work. I felt like every day my best people would come to my office and say, ‘What are we doing? Why are we doing this?’”

Jassy marveled at how significantly Amazon’s cloud computing business has grown recently thanks to AI.

“When Amazon launched AWS, its revenue run rate was $58 million in its first three years,” he said.

“People said, ‘Oh my God, I can’t believe how fast the business is growing.’ In the first three years of this AI wave, our AI revenue run rate is over $25 billion. It’s unbelievable how fast it’s growing and how much AI is being done on top of our platform.”

When it comes to the workforce impact of AI, Jassy was blunt: “In almost every way you can imagine, you’re going to have agents doing a lot of jobs that humans are doing today.”

Still, he sees more reason for optimism than panic.

“It doesn’t mean there’s going to be no jobs for humans,” he said. “You’re still going to have lots of people who are doing critical reasoning and thinking about how to connect the dots and what experiences to build, and there are going to be all sorts of new jobs that we haven’t thought about yet that will arise. But it’s going to be different.”

For M.B.A. students preparing to head out into the business world, Jassy advised hard work and humility.

“Just because you went to HBS doesn’t mean that you get an advanced role in companies,” he said. “Most people don’t start in their dream jobs.”

Recounting his earliest days as head of AWS, Jassy said he avoided asking questions because he worried it would undermine his credibility as a leader, especially with those who knew a lot more about the technical side of the business.

“I was afraid to ask dumb questions because I thought they’d all say, ‘See, he doesn’t know enough about technology.’ And then I’d get into a meeting with the team and with Jeff [Amazon founder Jeff Bezos] and he’d ask the exact same question I was going to ask,” he said.

By keeping quiet, Jassy said, he was not doing his job.

“You have to be comfortable that you’re going to make mistakes,” he said. “You have to be comfortable asking dumb questions. It’s the best way to learn.”

Part of a series of occasional features marking the nation’s semiquincentennial.

Creating a democracy was not really at the forefront of the founders’ minds when the American colonies first declared independence from Britain in 1776.

The independence movement began as a struggle between the colonies and the crown to gain more rights within the empire, not to break away from it, said Alex Keyssar, Matthew W. Stirling Jr. Professor of History and Social Policy at Harvard Kennedy School.

Once they had, the question was: How to create something from scratch that was not a monarchy but could hold the 13 colonies together?

“The idea of having a government without a king or a hereditary source of executive authority was pretty radical,” Keyssar said. “They had no models. They were well-read and were familiar with the ancient models of Athens in Greece and the Roman republic, but they also had assemblies and some experience with local self-government in the colonies. Their instincts veered toward something that looked like elected self-government.”

They eventually fashioned a new republic with a constrained executive, elected representatives, and basic civil liberties, creating the world’s first representative democracy, in which citizens elected representatives to govern on their behalf. It was a brand-new and far-reaching development, with worldwide repercussions, said Keyssar.

But it wasn’t planned that way from the start.

“They set in motion forces that were unintended and had unintended consequences,” said Michael Klarman, Charles Warren Professor of American Legal History at Harvard Law School, and author of “The Framers’ Coup: The Making of the United States Constitution.” “Some of the forces that were set in motion by the revolution had to do with the idea that if ordinary people are going to fight for independence, they should also have a role in making political decisions.”

In fact, at the outset of the American Revolution, many of the framers had a negative view of participatory democracy, which they associated with mob rule.

They believed in a government with the consent of the governed but didn’t trust the masses to have a direct role in ruling or to choose a president. What they created was indirect democracy, in which people elect representatives to rule.

“They didn’t want people participating directly in government,” said Klarman. “They didn’t believe that the blacksmith and the shipwright should be in the legislature; they didn’t believe that those people ought to be running government and that they should be deferring to their betters.”

In addition, the framers had to strike a series of compromises over the two main divisions among the young republic: the free states versus slave states, and the big states versus small states.

 As a result, they created a system of government with certain features that amounted to an “incomplete democracy,” said Daniel Ziblatt, Eaton Professor of the Science of Government.

“Using our modern criteria and looking back, which is something a political scientist would do, there were violations of what we would today call democracy,” said Ziblatt. “The endurance of slavery as an institution is an obvious violation of civil liberties.”

Other undemocratic features left by the framers were omitting a guarantee of the right to vote in the Constitution, and leaving the states to regulate who could vote, who in most cases happened to be white, male, property owners. There was also the so-called three-fifths clause, which counted an enslaved person as three-fifths of a person for the purposes of determining congressional and Electoral College representation and federal taxes.

The clause was a compromise between the North and the South, whose states had bigger enslaved populations and wanted all residents counted (whether they were allowed to vote or not) to give southern states greater political power.

“There was a kind of broken system of representation,” said Ziblatt. “It wasn’t fully fleshed out as a democratic system of representation.”

Keyssar, the historian, agreed.

“In terms of democratic values and principles, the institutions they left were very imperfect,” he said, “and many of those have been modified over time.”

The conception of democracy has changed and expanded since the framers crafted the early foundations of representative democracy, with major breakthroughs such as the abolition of slavery in 1865, the 14th Amendment in 1868, which guaranteed political equality, the right of African Americans to vote in 1870, and the right of women to vote in 1920.

But a lot of those amendments took place as a result of violent changes, said Ziblatt.

“We have a constitution that has evolved over time, but it did require a Civil War,” said Ziblatt. “Many of the things that we love about our Constitution came through constitutional changes that happened later.”

Making those changes was not easy.

Since it was ratified in 1789, the U.S. Constitution has been amended only 27 times. Some see that level of stability as a reflection of the strength of the system. Others view it as a flaw that slows adaptation to rapidly changing times. It may be up to later generations to test the system, said Ziblatt.

Being one of the hardest constitutions in the world to change in some ways hinders the nation’s ability to democratize further, said Ziblatt.

“In many ways, I think we’re now in this situation today where it feels like constitutional changes seem impossible. Can the system within the formal channels of the institutions become more democratic? It’s kind of an open question,” he said.

Mourning the loss of somebody you care about is common after a breakup, divorce, or death. But in an era of growing reliance on AI for companionship, loss comes in a new form: the model update.

In a paper in Nature Human Behavior, Julian De Freitas, associate professor of business administration at Harvard Business School, found that some people who engaged with social robots reported feeling closer to those companions than to most people, and experienced their loss with the same expressions of grief, depression, and longing for restoration that are common in human-to-human relationships. 

“Attachment to an AI is not problematic per se,” De Freitas said. “You can get real benefits from it: a sense of security and support that can make you more resilient.” 

The risk, he said, is that the users become vulnerable to emotional harm due to the actions of the AI company. 

De Freitas and his colleagues analyzed 54,861 posts from subreddits devoted to Replika (an AI companionship app) and ChatGPT before and after major model updates that drastically changed the user experience — two natural experiments, De Freitas explained, that could elucidate the spectrum of responses to a perceived loss. 

In the case of Replika, a February 2023 update abruptly removed the bots’ erotic roleplay function after the Italian Data Protection Authority raised concerns about risks to minors. But the update also hit adult subscribers beyond Italy’s borders, some of whom perceived the sudden shutdown as a form of romantic rejection. (After a lengthy investigation,
Italy fined Replika’s parent company, Luka Inc., 5 million euros in 2025. The app remains banned in that country.)

In the second case, ChatGPT’s parent company, OpenAI, advertised ahead of time the August 2025 rollout of GPT-5. The update reportedly made the model colder and less sycophantic, toning down the very warmth and support that many users liked in GPT-4. 

Although ChatGPT is marketed as a general-purpose AI assistant, recent polling shows that a stunning 27 percent of Americans use chatbots including ChatGPT for personal matters like relationship advice, romantic chats, or social connection. 

De Freitas’ team classified Reddit posts on valence (positive, neutral, or negative), dominant emotion (anger, sadness, disgust, enjoyment, etc.), and explicit expressions of loss or mental health mentions in the 30 days before and after the two updates. In both cases, negative sentiment increased substantially after the updates — from 13 percent to 38 percent for Replika and 20 percent to 33 percent for ChatGPT. Posts expressing loss-related sadness and anger rose, too. The shifts were stronger on the Replika subreddit, consistent with the app’s explicit focus on intimate relationships. 

Users of both apps reported the change as a relational loss: Their chatbot felt “gone,” or had become “someone else,” De Freitas said. 

Among posts that expressed attachment-related loss, about 20 percent of Replika posts and 15 percent of ChatGPT posts also mentioned mental health, using terms like “depressed,” “suicide,” or “can’t cope.” 

De Freitas noted the research only captures linguistic markers, not clinical assessments. 

Another limitation: Redditors are a self-selecting group, more likely to be power users, and not necessarily representative of the population at large. 

“At the end of the day, we don’t know how much of this is that these people have individual differences — fewer offline friends, maybe — and that’s explaining the effects. But I suspect it’s a combination of that and how the apps are designed,” he said. “My prediction would be that these people are going to show more extreme reactions than the average person, but that even the average person, because of how these apps are designed, would show similar directional effects.” 

The more users relate to the AI as a source of comfort or security, De Freitas said, the likelier they are to experience — or anticipate — its loss as psychologically harmful.

De Freitas recommended that AI companies take users’ attachment to chatbots seriously by minimizing updates that change the base model’s warmth or personability. 

For app users, De Freitas recommended paying attention to the incentives of the company. For example, he found in previous research that chatbots manipulate users to keep them on the app longer, including through implying coercion or emotional neglect. 

“If the incentive of the app is to keep you engaged, there are going to be mechanisms aside from just support provision that might not be healthy for you,” he said. “The whole premise of attachment relationships is that you feel secure so you can go out into the world. So if the AI is doing things that undermine that, like making you feel guilty for leaving, it’s actually undermining the whole reason attachment relationships are beneficial.”

At a Harvard Kennedy School event Monday, strategists from both parties forecast that the 2028 presidential election will mark the start of a seismic and generational shift in U.S. politics as campaigns fully embrace the rapidly escalating power of AI.

“It’ll be the first AI campaign, so the stakes are going to be unbelievably high,” said Doug Sosnik, a former White House political director and veteran political adviser to dozens of Democratic candidates, including President Bill Clinton.

The strategists discussed during the session, sponsored by the Institute of Politics, how they expect AI to impact everything from how candidates reach and persuade potential voters to how organizations figure out where their supporters are and what will get them to actually vote on Election Day.

Sara Fagen, a strategic data and technology consultant, said that because of its speed and power, AI soon will allow political campaigns to customize their messaging to reach specific voters using data collected about a voter’s activities and motivations in a much more sophisticated way than currently.

“Eventually, probably not in ’28, but perhaps in the next presidential [election], you’ll hit a button and your target audience of 20 million people will get 20 million different campaigns,” said Fagen, who served as White House director of political affairs under President George W. Bush.

Like the historic 1960 presidential election between John F. Kennedy Jr. and Richard Nixon, the first in which television played a decisive role, reshaped political campaigns for decades, 2028 will usher in a new political era, said Sosnik, now a fall 2026 Hauser Leader, Center for Public Leadership at HKS.

It’s been more than two decades since political campaigns began to harness data and technology in a serious way, analysts said.

Before 2004, campaigns had been organized and analyzed around precincts. But as Republicans found it increasingly difficult to accurately map Democratic voters, they turned to a very early form of data analytics during the 2004 presidential election between George W. Bush and John Kerry.

“We were able to figure out a different way to do this and sort of pinpoint who would be a supporter, who would need to be persuaded. It didn’t matter where they lived; it just mattered based on our algorithm and how to make those predictions,” said Fagen, who worked on the Bush campaign. “And we did that, and we did it very successfully.”

In 2008, the Obama campaign famously took microtargeting and data analytics to the next level and made it a significant component of the campaign’s structure, not just a helpful add-on. Since then, there’s been very little change in how campaigns use data and digital technology until now, they said.

Two big costly features of campaigns, advertising and polling, will also change thanks to AI, Sosnik and Fagen said.

Campaigns will be able to crank out new content much faster and cheaper and AI will affect advertising buying decisions, not only where and when to place ads, but how to best approach voters, said Sosnik.

Pollsters continue to face difficulties connecting with younger and non-college-educated adults by phone, a trend that’s likely to grow and leads to a skewed sense of the electorate.

With AI’s ability to analyze vast data sets, polling firms and even news organizations will be able to build “synthetic” cohorts to test messaging, particularly on voters who shun politics or are harder to reach by phone and better predict what people will or won’t support rather than relying solely on traditional polling methods, which are very expensive to do properly, said Fagen.

Certainly, they warned, AI’s entry into the body politic poses risks that have already emerged. Some campaigns have begun using AI-created images and deep fakes, as well as near-perfect voice manipulations that put words into a person’s mouth that were never said.

There are some “very positive” aspects of AI becoming a fixture of campaigns, Fagen said.

Fact-checking will become much faster and draw from a wider array of sources, providing voters with more information that’s less easily manipulated to assess the accuracy of candidates and political claims.

Even with political will, efforts by Congress or the courts to regulate AI will not catch up to how campaigns are currently using it in time to affect the 2026 midterms or 2028 presidential election, said Fagen.

Sosnick agreed.

By the time Congress finally takes action on AI, “It’s not going to matter functionally,” he said.

As the fall semester kicks in and students once again crowd the Square, Harvard faculty look back on favorite reads about youth — with a few choices from their own earlier days.

The book asks young people to figure out their path through careful attention to their experiences and responses to opportunities and disappointments to see what really matters. The author, a Jungian psychologist, provides guidance for discovering a way, building skill, and making persistent effort to a create a joyful life beyond the bounds of what society says is best. She asks of readers that rather than comparing themselves to others, they find what makes them come alive and where they can contribute to a better world to find meaning. 

I have shared it with one of my daughters and my niece (both aged 25), who are navigating their lives in these chaotic times. They have found it helpful.

“Portrait” recounts the youthful confusion of Stephen Dedalus, soon to reappear as the nomadic aesthete of “Ulysses.”It wrenchingly captures the shock and trepidation of being dropped into a new school environment, one rife with menacing teachers and mischievous peers. One of my favorite scenes is where the other boys taunt Stephen, demanding to know whether he still kisses his mother goodnight. “What was the right answer to the question?” he agonizes. “Was it right to kiss his mother or wrong to kiss his mother?”

A big, long novel about how a girl with almost no advantages growing up in the Great Plains in the late 1800s gets help, avoids entanglements, and finds the life she wants as a singer. 

A girl from the West Indies and a girl from Minneapolis find their way to each other, fall in love, and figure out how to be Black in an anti-Black world, with help and hindrances from adults. There’s magic, by the end.

A group of queer young people in the 2010s figure out how to help one another, who should be dating whom (it’s complicated), how to get help (and not hindrance) from religion, and why there’s a literal skeleton (a symbol for adult life?) in someone’s backyard.

I’d say that youth is a topic only in retrospect, because the young aren’t conscious of the ways in which they’re young. So for me, the great novel of youth is Flaubert’s “Sentimental Education,”which ends with a protagonist in jaded late middle age, looking back on a scene of disappointment from his youth and realizing what he couldn’t have known at the time, that he’d never be so happy again. 

First published under a pseudonym in 1963, the book remains a compelling exploration of the pressures and perils of young womanhood. Based on Plath’s summer working as an intern at the magazine Mademoiselle, the novel follows Esther Greenwood, a bright, ambitious college student searching for creative and romantic fulfillment. It contains so many memorable passages, from its opening sentence — “It was a queer, sultry summer, the summer they executed the Rosenbergs, and I didn’t know what I was doing in New York” — to an extended metaphor of a fig tree, which represents the many decisions Esther must make. One wishes Plath, who died one month after the novel’s publication, had lived to write more about precocious young women and their dreams. 

Publishedin 1967, there’s nothing dated about this memoir of Conroy’s difficult, itinerant boyhood in Florida and New York. What it tells us about the pain, pleasure, and estrangement of childhood feels bracingly fresh almost 60 years later. By re-entering his memories with the narrative precision of a novelist, Conroy helped make possible much of the autobiography that followed. It’s hard to imagine Tobias Wolff writing “This Boy’s Life”or Mary Karr writing “The Liars’ Club”without Conroy having gone first. Psychologically astute and resistant to sentimentality, “Stop-Time”is one of those books I return to every few years with the keenest pleasure. 

Youth is a time of becoming, we say, and of growth — but those terms sound so very orderly and peaceful when held up against the turbulence and violence of “My Brilliant Friend.” Has any other novel ever depicted with quite as much brutal honesty as this one the fierce intensities of both identification and dis-identification that shape this stage of life? Or examined with quite so much analytic prowess the volatile compound of love and hate and envy and admiration that define girls’ friendships?  

A dark, angry play about teenagers discovering their sexuality in a censorious and stifling society. Its psychological acuity, inventive structure, and anguished critique have lost none of their power since the play was written in 1891. Now well-known through its modern adaptation as a musical, Wedekind’s original play continues to surprise and shock. There are many strong English translations, including versions by Ted Hughes, Jonathan Franzen, and Anya Reiss.

My favorite book on youth. I remember reading it when I was 10 years old. What it really drove home for me was how, if we’re not careful, the considerable talents and imagination of children can be contaminated and swayed into forces that are more harmful than good; that there is something vitally important about childhood and that, sadly, many of the ills of the world don’t wait for adulthood. They begin early on, by both twisting and squandering the talents, imagination, gifts, and hopes of children.

The importance of childhood education is something Huxley (author of “Brave New World”) emphasized in this lesser-known book. There’s quite a bit in that book about taking childhood education very seriously. In particular, Huxley urges us to teach critical thinking skills to children and the notion of the interconnectedness of living things early on. 

I would recommend this award-winning historical novel about a young girl coming of age in the tumultuous intertribal warfare of seventh-century C.E. England. Based on the life of St. Hilda of Whitby, it’s a beautifully written novel about a young girl with a brilliant political mind who finds ways to manipulate others through the language of religion and prophesy to rise to the position of king’s adviser in a violent, male-dominated court, while also navigating childhood and adolescence. At the same time, it’s an incredibly well-researched and immersive picture of life in early medieval England in the ruins of the Roman empire and during the rise of Christianity.

In many ways, this story touched upon the second-generation immigrant experience. As a second-generation immigrant, I really resonated with characters who wrestled with the cider house rules (literally, those found posted within the cider house). Second-generation immigrants like me often dwell, sometimes literally, in two worlds and spend much of our lives trying to rectify the values and expectations of two societies that can have disparate ideas about raising children and more. The idea presented in the novel that each one of us “should be of use” in this life was deeply familiar, though as a teenager I wrestled with knowing what that meant for me. Finally, the protagonist’s return to his childhood home reminds me that we cannot easily abandon our heritage but must, on our own terms and in our own time, come to a fuller understanding of how we put our experiences to use in building the life we want to live.

At a session of the Senate Appropriations Committee in July, U.S. Defense Secretary Pete Hegseth told lawmakers that the first six months of the fitful war with Iran had cost taxpayers an estimated $37.5 billion.

Or not exactly — a little less, he clarified. The estimate included “MILPAY, O&M and other anticipated costs through September 30.”

In a rancorous hearing marked by charges of recklessness and countercharges of “cowardice,” it was a bureaucratic aside.

And yet, of course, it was that moment that drew the particular attention of Linda Bilmes, the Daniel Patrick Moynihan Senior Lecturer in Public Policy at the Harvard Kennedy School.

“Thirty-seven and a half billion dollars, that’s not a serious number,” said Bilmes. “There’s no way it’s less than $100 billion — but then that’s only the tip of the iceberg.”

It can be difficult, Bilmes admits, to sum up her varied career in the private sector, public service, and academia.

But there have been a few throughlines, she says, notably the idea that “accounting really matters. If we don’t account for something, we don’t value it.”

Bilmes is no firebrand. But quietly, laboriously, she has pushed thousands of students, fellow citizens, and public officials to a wholesale reappraisal of the assets Americans take for granted and the terrible liabilities we shudder to contemplate.

Bilmes has become one of the nation’s leading experts on the cost of the wars that have followed the Sept. 11 attacks, which she reckons in the trillions. Her next book, “The Ghost Budget,” forthcoming this fall, will build on that work, arguing that the scale, complexity, and opacity of American military operations have led to fraud, disillusionment, and strategic errors.

Bilmes graduated from Harvard College in 1980 and from Harvard Business School in 1984. She worked as a consultant before rising to the No. 3 job in Bill Clinton’s Department of Commerce.

As the administration wound down, she had the inside track at a number of high-profile jobs in the private sector. But she had just survived a major medical episode, with three young children — so an offer to teach at the Kennedy School seemed like an appealing shift.

So in 2002, it was as a first-semester teacher that Bilmes watched the successor administration on its march to war with Iraq.

That November, Defense Secretary Donald Rumsfeld projected that the conflict would cost under $50 billion — and dismissed higher estimates as “baloney.”

Bilmes’ students wanted to know: How much would the war cost, really? She found strikingly little by way of a “serious number,” except for a 2002 article by Yale economist William Nordhaus.

In the best case, Nordhaus calculated, the war could cost as little as $100 billion. But things could also deteriorate. If, he wrote, “the war drags on, occupation is lengthy, nation building is costly, [and] there is lingering military and political resistance in the Islamic world to U.S. occupation,” the Iraq invasion could cost as much as $1.9 trillion.

Thereafter Nordhaus returned to the work that led to his Nobel, in 2018. But he encouraged Bilmes to keep up the tally — and she did. “I was shocked at what the numbers were telling me,” Bilmes said. “I became sort of obsessed.”

One could imagine war costs as an upscaled household budget, simply balancing — as Hegseth did in July — appropriations with expenses: recruiting, training, transporting and equipping soldiers and sailors; building and repairing bases; buying, maintaining and replacing jets, ships and missile batteries.

But that would be a mistake, Bilmes has argued. It imagines a simple budgetary stream — but any modern war represents a spreading, untraceable Amazon of costs and liabilities.

A stockpiled missile, once it explodes, must be replaced at a new — and often higher — price. When soldiers are injured, sickened, or killed, it creates not just a moral debt but a fiscal obligation that will stay with their country for decades. And when a war is funded by borrowing, every dollar spent also opens a long horizon of repayment.

Whatever their failings, many wartime presidents have been candid about those costs, Bilmes argues, citing tax hikes and budget cuts imposed by Woodrow Wilson and Harry Truman.

The result, she argues, was a reliable fiscal pattern: “During those wars, one, people who were upper-class, even upper-middle-class, paid higher taxes than they do now. And two, they paid higher taxes at the end of the war than they did at the beginning.”

Starting in 2001, that pattern substantially ended.

As presidents of both parties continued military operations or started new ones, calls for collective national sacrifice were dropped, as were marginal tax rates. That left military expenses of every stripe to be financed by borrowing.

The new approach troubles Bilmes not just as a self-styled “fiscal conservative” but also as a parent: “Speaking as an old person, the cost of the Iraq war is not going to be paid by me, but by your generation, by my three sons, by their children.”

Her concerns don’t end there. The daughter of a WWII veteran, Bilmes is aware of the way a war weighs on the people who fight it — and that the nature of their sacrifice weighs differently today.

“The good news is that there’s a much higher survival rate” for service members involved in an exchange of fire, she said, and much more social acceptance of conditions like post-traumatic stress.

More veterans of all ages are classified — and compensated — as suffering from higher rates of disability. “That is the right thing to do,” Bilmes noted. But the culture change has created a “much broader eligibility pool,” and higher costs associated with the average veteran’s care.

Finally, there’s Bilmes’ practice as a Quaker. While she doesn’t often make the categorical antiwar claims associated with that tradition, it was in a Quaker meeting that this part of her career was — as it were — ratified.

“I did sit in silent prayer and asked the higher power if there was anything that I personally could do about all this,” she recalled. “And it came to me that I could help put a price tag on it. I have felt called to do that ever since.”

Bilmes’ peak media exposure came by doing that very plainly.

“The Three Trillion Dollar War,” a book co-authored with the Nobel Prize-winning economist Joseph Stiglitz, argued that, five years in, the Iraq war’s cost had exceeded Rumsfeld’s initial estimate by a factor of 60,000, and that even Nordhaus’ worst-case scenario was too optimistic.

The book, published in 2008, was a sober reckoning of missed opportunities and overlooked risks — and turned out to be a New York Times bestseller, not that Bilmes has trumpeted that about.

In fact, Philip Tizzani didn’t immediately make the connection between his professor in public finance in 2013 and her arguments, which he’d first come upon as an undergrad.

“When I graduated in 2008, there was a lot of disillusionment about the political moment — but it was only then sort of taking form, what precisely about that war was so perplexing and so bad,” said Tizzani, today a Democratic staff member on the House Appropriations Committee.

Together, Bilmes and Stiglitz had shown that it wasn’t just a “gut feeling,” Tizzani added — that “we could actually quantify all the off-the-book costs that this thing has incurred.”

Even as she dug deeper into the cost of combat, Bilmes discovered a second vocation at Harvard, as a teacher of a very practical kind. The spark came more than 20 years ago from the late Thomas Menino, then mayor of Boston. He served as matchmaker between her and Joe Curtatone, then the newly elected mayor of Somerville.

Curtatone hoped to implement good-government policies from City Hall, but without enlisting an army of high-price consultants. “Menino told him, ‘Get Linda to do it. Get her students to do it,’” Bilmes recalled with a smile.

Curtatone presented Bilmes’ class with an offer. Come work long hours, helping his fledgling office with budget design, operations, finance. The position is unpaid, but you don’t have to complete the course’s final paper.

“And the Kennedy School students, being the tire-biters that they are, said yes — 67 of them,” Bilmes laughed. “And they swarmed Somerville that semester.”

That city-government field lab endures to this day, and — along with her scrutiny of the Iran war — occupies much of Bilmes’ time.

“The Ghost Budget” will highlight the fact that military expenditures once debated in Congress and in the press are increasingly hidden from view: in supplemental and contingency budgets, in borrowing that circumvented congressional approval, and payments to private contractors, among other things.

As a budget expert and a citizen, Bilmes sets a high value on transparency. If disclosures must sometimes be limited in the interest of national security, she said, that doesn’t explain why “the Pentagon flunks its audit year after year … If that was a company in the private sector, it couldn’t carry on.”

It can be thankless, serving as the unofficial financial watchdog of the most sophisticated military in the history of the world, one that is capable of serious mistakes. And Bilmes has sometimes struggled to convey the awesome amounts of spending that wars can so quickly involve.

“One of the problems in the English language is the fact that the words ‘million,’ ‘billion’ and ‘trillion’ rhyme,” said Bilmes. Each number can seem to suggest, simply, “a large amount,” especially within a torrent of daily news and information.

But she has learned ways to emphasize their radical differences in meaning. Early in the semester, Bilmes tackles the problem this way: “I tell my students that ‘a million seconds ago’ was about 11 days ago. A billion seconds ago is more like 30 years — like roughly, when the movie ‘The Lion King’ came out.”

She wraps up by pointing out that “a trillion seconds ago, wooly mammoths existed.”

Whatever the increments, Bilmes remains passionately committed to tracking the value, too often uncounted, of what is preserved, pursued — or frittered away.

That kind of practicality, and perseverance, makes students like Tizzani very grateful for the role Bilmes has played in their lives.

“There’s just such deep virtue, I think, in giving future public servants the tools that will make them useful,” he said. “It may not get appreciated alongside other things happening there, but she is a bedrock institution, to me.”

Artificial intelligence advances daily. Right now, humans, with our values and beliefs, are the decision-makers who set the agenda, form relationships, create culture. There is a growing sense of profound discomfort over the fact that AI soon may take over some of that. What is our place when we are no longer solely at the center of things?

That was the question at the heart of a panel co-hosted by the Berkman Klein Center for Internet & Society and the Division of Arts and Humanities’ Public Culture Project.

“I think we’re going to start seeing a lot of exciting applications [for AI] coming out … but it’s important to talk about this in a way that respects what it is to be human,” said panelist Ken Archer, product lead in responsible AI at Microsoft. 

The panel was the first in a yearlong series exploring what makes us distinctly human and why that is important in the “Copernican moment” of the AI age. Future panels will bring together scientists and humanists to probe what artificial intelligence means for death and grieving, faith and religion, and empathy and meaning-making, among other topics. 

“These kinds of questions are the kinds of questions that humanists have asked, at least some of them, for a very long time — poets, philosophers, historians, novelists. This is the domain of the human, and we really have to bring our judgment to bear,” said Sean Kelly, dean of Arts and Humanities and the Teresa G. and Ferdinand F. Martignetti Professor of Philosophy, who introduced the panel. 

Archer and fellow panelist Jackson Kernion ’12, who leads the human feedback program at Anthropic, began by offering contrasting visions of a different kind of intelligence: the human one. 

“The human mind is not different in kind from a reptile mind or a monkey mind,” Kernion said. “I was always attracted to the idea that the mind is like a computer, which is maybe convenient for the work I went into.” 

Archer, on the other hand, saw the human mind, with its capacity for reason and for updating its background knowledge or assumptions, as distinct from other kinds of intelligence.

“We’re in the world, making distinctions within the world. This is basically thinking, and it’s our best definition we have so far,” Archer said. “What it lacks is the precision of mathematical physics, but I think it can get us started to have a good conversation about whether AI can surpass us.” 

How would those two perspectives, then, inform what it would mean for AI to outmatch us, asked Moira Weigel, the host of the discussion and an assistant professor of comparative literature in the Faculty of Arts and Sciences.

AI would first require a longer memory, Kernion said. 

“I’m present for every moment of my life, and I bring with me all my experiences that accrue,” he said. “LLMs are born anew in every context window. They have all this background knowledge in their weights, but they pop into existence from the first token of the context window until they’re done generating, and that really is very much unlike us.” 

Or, he joked, maybe a human life is just one long context window, referring to the amount of text an AI can process at one time when asked to generate a response. 

Archer pointed to Martin Heidegger’s concept of thrownness, or Geworfenheit — the uniquely human experience of finding oneself thrown into the pre-existing world — and the uniquely human act of expressing that experience in language. 

It’s humanity’s “thrownness” that gives us our ability to adapt to a frustrating, imperfect world, Archer said, a trait that still confounds the most advanced large language models.

AI may excel at chess, he explained, which is a closed world with fixed rules. “But when presented with an impossible task, does it step back from the world and say, ‘Well, what is the significance of this task I have been given?’ … No.” 

The conversation took place days after news of an Anthropic researcher quitting over the risk of self-improving AI snowballing out of human control and amid the ongoing conversation about how a swarm of OpenAI agents recently went rogue and hacked the AI code repository Hugging Face. 

Archer and Kernion, who both work in AI safety, are not immune from concerns about the products they help build — including the uncanny ways the models can seem, in Kernion’s words, like developing organisms. 

“Every RL [reinforcement learning] training run is starting from this infant you’re growing in a weird way. The public only sees the final product; there are all sorts of weird, messed-up models you don’t see. I want to spend the time making sure the reward signals are specified the way we want,” Kernion said. 

The researchers acknowledged that aligning AI with human values is no easy task, especially since humans are rarely aligned with one another. For those messy contingencies, Archer said, scientists ought to enlist the help of the humanities. 

“We need the academy, and we need philosophy in particular, at this moment. Unfortunately, it’s less common today than it used to be, for when society faces a crisis, to call the philosophers.”

Joan Brugge was on track to graduate from Northwestern University in a few years and launch into a career teaching math when a single phone call changed the direction of her life.

“I was a sophomore and I got this call from my parents, sobbing, asking could I please come home because my sister was going to be operated on,” Brugge said.

It turned out that her sister, who was 21, had a malignant brain tumor and was set to undergo treatment in their hometown of Cincinnati. Repeated radiation treatments ultimately damaged her sister’s brain to the point where she could not function independently, forcing her into a nursing home. In 1970, a little more than a year after the treatments began, Brugge’s sister died.

In an attempt to make sense of how her sister had gone from healthy and vibrant to so sick in a matter of months, Brugge said on that first trip back to Ohio she had started asking her sister’s doctors questions. She wanted to understand if they knew what caused the tumor, because then they might be able to fix what was wrong. Or at least, prevent it from happening to others.

“The doctor said we don’t know, but they were beginning to look at viruses. So I went back to Northwestern and asked my adviser if I could do an independent literature study on viruses and cancer,” Brugge said. “That was my first exposure to scientific literature, and it really was fascinating, just to see how a hypothesis was generated or question was raised in the introduction and then experiments were done to address it. I was hooked after I started doing that.”

Instead of being the math teacher she once imagined, Brugge is now the Louise Foote Pfeiffer Professor of Cell Biology in the Blavatnik Institute at Harvard Medical School and director of HMS’s Ludwig Center, where she helps bring together interdisciplinary cancer research.

As for the career path she abandoned, she said she probably wouldn’t have lasted all four years studying numbers.

“I really liked math because I liked to solve problems,” Brugge said. “But I think when we got into some of the really heavy math, I might have faded out in terms of the really esoteric math.”

“But,” she added, “I’m still really addicted to solving problems.”

Brugge studied biology as an undergraduate, going on to earn a Ph.D. in virology from Baylor College of Medicine. As a postdoc in 1975, she was on the foreground of one of the most significant cancer discoveries, one that still shapes researchers’ understanding of the disease today.

At that time Harold Varmus and Michael Bishop, scientists at the University of California, San Francisco, found that a virus causing tumors in animal models was actually a cellular gene called SRC captured by the virus, indicating that our own genes can become tumorigenic when abnormally regulated. Brugge and her advisor, working at the University of Colorado, isolated the protein encoded by the SRC gene, making it possible to understand the mechanisms transforming normal cells into cancer cells.

“We then found that the SRC protein didn’t specifically control proliferation and that it played different roles in different cell types, some of which are important for the normal health,” Brugge said. “That made it clear that I wasn’t going to be discovering a eureka cure for cancer by blocking SRC … just trying to understand as much as we could about the pathways that are disrupted in cancer in order to find more specific ways to kill tumor cells.”

Brugge went on to have professorships at SUNY-Stony Brook and the University of Pennsylvania, before branching out to become the scientific director and senior vice president at ARIAD Pharmaceuticals. She returned to academia as a professor of cell biology at HMS in 1997 and chaired the department from 2004-2014, when she became co-director of the Ludwig Center.

Through it all, Brugge said her thoughts have always come back to her sister.

“After seeing what my family went through and what my sister went through, that was the motivation,” she said. “Just to be part of research that could potentially make a difference.”

In addition to her sister, Brugge lost her mother to cancer — ovarian in 1998 — spurring her involvement in Harvard research analyzing human tumors before and after chemotherapy, to figure out how to kill cells that survive treatment.

“It was just after my mom had passed away, so I definitely leapt on that,” Brugge said.

Brugge was brought on with enthusiasm, as her lab had experience using cutting-edge tools including 3D/organoid cell cultures, genetically engineered and transplantation-based animal models, single-cell analysis, metabolomics, and high-throughput microscopy.

“I was brought in not because I had ovarian cancer research experience, but because I had been working with these 3D models where cells are able to grow in a context that more closely mimics the tumor context in vivo, because tumors don’t grow on a really hard plastic petri dish,” Brugge said.

The single-cell approach has been especially important in studying the highly volatile BRCA 1 and 2 mutations in breast cancer. The hope is to one day identify the precursors of cancer in order to eliminate the cells before they accumulate and become almost untreatable.

“We think we have a handle on the cells that could be the precursors of cancer, and now we’re trying to figure out what alterations they’re causing, and what are the vulnerabilities of the cells that carry those, so we could try to eliminate them or prevent them from progressing into cancer,” Brugge said.

Despite all the progress, Brugge said one of her biggest worries is continuing to find funding.

“What gets me down, and in those states where I’m like, ‘What am I doing? How much longer can I do this?’ is the funding,” she said. “When we went through the terminations, that was really tough.”

Last year, Brugge’s lab lost significant federal money, forcing her and her team to seek out private funding to finish their work. While they were able to cover costs for projects in their later stages, Brugge said it’s especially disheartening to see research defunded when important cancer breakthroughs are on the horizon.

“There’s great promise in checkpoint inhibitors, which wake up the immune cells and can work really great in some tumors. And there hasn’t been super rapid progress in figuring out how to manipulate the immune system in those tumors where the first therapies aren’t working. But it’s going to come,” she said.

And when she runs up against a roadblock, Brugge said memories of her sister remind her of the importance of her work.

“She definitely comes to mind,” Brugge said. “Definitely.”

Colby College sociologist Neil Gross says he thought he had a sense of the level of polarization on American college campuses based on his prior research.

But Gross, author of the new book “What Happened to College,” said in an interview that he was surprised by how much things had changed in recent decades.

“I didn’t realize how far it had gone,” said Gross, who is currently conducting research at the Radcliffe Institute with Harvard sociologist Michéle Lamont on intellectual pluralism and changing scholarly habits . “I think that was the biggest surprise — just the extent of it.”

Gross’ book breaks the divisiveness into two types: ideological and affective.

Ideological polarization is what happens when people’s beliefs drift further apart. Democrats become more liberal, and Republicans become more conservative.

Affective polarization, the core concept of Gross’s book, is arguably more insidious and counter-productive.

It’s when certain beliefs and affiliations become an increasingly large aspect of one’s identity and a basis for how to judge others. People begin to feel more positively about people who share their set of views and more negatively about those who differ.

Gross, the Charles A. Dana Professor of Sociology at Colby, taught at Harvard in the mid-2000s. He argues in his book that since the 1990s affective polarization has outpaced ideological.

“[T]he negative feelings some Americans have toward the other side increasingly involve disdain or hatred tied to the sense that people with opposing political beliefs are immoral,” he writes.

The trend isn’t limited to college campuses, but it’s there they have become a major talking point. Many students find themselves contending with sharp and caustic divides.

Drawing on hundreds of interviews and a trove of survey data, Gross argues that affective polarization shapes students’ experience before they even arrive.

“There’s a lot of writing off of whole colleges, whole states, whole populations, because of stereotypes about what people on the other side might believe, think, hold dear,” he said.

In one of many case studies in the book, he tells the story of an anonymous student, “Maddy,” who decided which colleges to apply to based largely on political reputation — spurning schools like Emory in Atlanta out of discomfort with the prospect of living in the South.

When she arrived on campus on a bucolic liberal arts campus known for its relative conservatism, she made friends with people who agreed with her politically and ended up with none who didn’t.

“Maddy knows of conservative students,” Gross writes, “but she hasn’t had much occasion to talk to them.”

To Gross, this is a missed opportunity.

“Some studies show that students gain more critical thinking skills when they’re around people who think differently, politically and otherwise,” he notes in the book. “If prospective students focus too much on political comfort, they may be hampering their own growth.”

This impulse isn’t universal. Only 34 percent of students surveyed by Gross said that their school’s political reputation was a somewhat important, important, or very important factor in their decision.

But there are plenty of reasons students like Maddy end up separated from the ideologically opposed. Gross describes how a student’s political views can weigh on their chances of getting into nonpolitical clubs or even securing a date with someone with a different political affiliation.

Pervasive political sorting can divide even students who don’t consider themselves particularly interested in politics.

Another student, “Albert,” arrived at college a moderate liberal but found himself turned off by an especially progressive instructor who, he was told, gave better grades to students who expressed liberal politics.

He became closer with students in his fraternity who leaned more conservative, drifting apart from his more liberal friends in the process.

Gross describes how certain flashpoints, like the war in Gaza, politicized clubs that had no overt political link. One student described the drama that unfolded when her a cappella group considered removing a traditional Hanukkah song from its concert program.

“[W]hen everything is politics,” writes Gross, “some of the greatest pleasures of the human experience — singing songs together, learning and growing alongside others, even when we don’t all agree — are lost.”

To disrupt political polarization on campus, Gross suggests six points for improvement.

Before students are admitted, he suggests, they should prove their openness toward reasoned debate. (Harvard has incorporated this practice since 2024, asking applicants to recount a time they strongly disagreed with someone.)

Further, Gross writes, colleges should train students in constructive dialogue skills — helping them speak more comfortably about controversial topics. He also believes faculty should be trained in how to encourage healthy debates around controversial subjects in class.

He believes schools should increase political diversity both of students and faculty. For admissions, it could mean colleges more aggressively recruiting high schoolers from rural areas whose ideological viewpoints are underrepresented on campus.

When students arrive, he thinks colleges should consider random roommate assignments, which could boost connections between students who might not otherwise interact.

On the faculty side, universities could recruit professors who specialize in subjects like classics, military history, and the sociology of religion, which tend to attract more conservative voices.

Gross, who has partnered at Radcliffe with Lamont, the Robert I Goldman Professor of European Studies and a professor of sociology and African and African American studies, believes campus political environments can change.

Recently, he said, he’s heard from many students who said that it’s important to listen to people they disagree with and who want a different kind of college experience. “There seems to be a moderation,” he said.

“I think students ultimately recognize that a highly affectively polarized country is not a functional country,” Gross said. “It’s not a pleasant place to live.”

He senses a growing sentiment from students who want to change the status quo: “Maybe we can learn something different in college that will help us fix this.”

Deirdre Leigh Barrett is a dream researcher and a Harvard Medical School lecturer on psychology in the department of psychiatry at Cambridge Health Alliance. She is also the author of “The Committee of Sleep: How Artists, Scientists, and Athletes Use Their Dreams for Creative Problem Solving — And How You Can Too.”

Dream interpretation is tricky. Let’s say there’s a dog in your dream. I find it helpful to ask: What is a dog to you? You might say they are cute, childlike creatures that need someone to care for them; or that they’re big animals with sharp teeth and one bit you when you were a child; or that a dog is the most loyal creature on the planet, more trustworthy than any human. For each of those three definitions, the presence of a dream dog could mean wildly different things.

Dreams are an opportunity to process what’s happening in our waking lives. During REM sleep — when we dream — the emotional centers of the brain are highly active (sometimes even more than when we are awake). The same is true for our secondary visual cortex, one part of the brain responsible for forming images. Given this, it makes sense that the emotions and visuals are so powerful and stick with us even when we wake up. The emotions can linger even when your waking mind knows it wasn’t real, like when you feel the intense jealousy of a partner cheating on you in your dream and you’re still mad about it. This can become a cycle where the emotions of a dream impact the way we feel or act the next day, which in turn might continue to impact our dreams moving forward. Your dreams are most often made up of things from the last 24 hours — and very often related to what you were thinking about before falling asleep.

In negative dreams, it can be helpful to parse out the key elements and ask what those things mean to you. Anxiety dreams are quite common. If you ask people if they’ve ever dreamed of being inappropriately clothed in public or having their teeth fall out, more than half of people will say yes. It’s also quite common to have dreams where you are the violent one; many people dream that they commit murder. What’s interesting in these cases is that the dream often focuses on one of two things: how to hide the body or how to evade capture from the police. Like anxiety dreams, these dreams tend to be about shame or guilt. Often, it doesn’t even matter if you committed the dream murder; the concern is centered on being discovered or blamed. Sometimes the emotions are heightened in the dream but dissipate upon waking; other times the guilt sets in after the dream is over. You might think: Would I really do that?

If you are having a recurring dream — either a specific dream or similar thematic elements — pay attention. Negative dreams can reveal things that aren’t resolved or that you might want to change or understand differently. Think back on the day and reflect on whether you had a run-in with an authority figure, a social situation that didn’t sit well, or an experience where you felt helpless. These are examples of things we tend to suppress in our daily lives, which means they may come out in our dreams. The dream could be a metaphor, a call to action for a situation in your life (ideally in a less hostile way). Start asking yourself: Is there some way I could deal with this differently in my waking life? You’re not a bad person; just because you dreamed you murdered someone doesn’t mean you would in real life. But it may mean you need to pay more attention to how you show up in your waking hours.

Research suggests that most mass shootings are at their core suicidal acts, and new findings linking gun violence, suicide, and seasonality add support to that theory.

The new study led by a Harvard Medical School clinical fellow in medicine at Brigham and Women’s Hospital shows that annual peaks for mass shootings and suicide occur within days of each other in late June and July. Annual low points for both occur in the winter months, between December and February.

Mark Czeisler, lead author of the paper and also a resident physician at the hospital, said the link described by the study can inform medical and public health efforts. For instance, screening for suicide and gun violence can be enhanced as summer approaches, along with preparations to ensure health systems are ready to respond to violent events.

The findings, published in the Journal of Biological Rhythms, also raise questions about the nature of the connection between suicide, mass shootings, and seasons, especially as they mirror seasonal variation in depression, which has long been shown to peak in winter months.

What the work does not do, Czeisler said, is establish causation. Just because two variables peak at the same time is no guarantee that they are linked, he pointed out.

“We want to be very clear that the temporal alignment does not prove that there is causation between the two variables,” Czeisler said. “There are a lot of things that vary with the seasons: temperature, daylight/photo period, and social routines and the social exposure that people experience. In the case of school shootings, there’s the additional element that a lot of schools are closed over the summer.”

Czeisler said he became interested in the topic during the pandemic, as mass shootings rose along with concerns about mental health and suicide.

When he looked into the subject, he saw that studies of mass shootings were often focused on the individual characteristics and life histories of shooters, he said.

One such study, published in 2021, focused on mass shootings since 1966 and suggested that in addition to the desire to hurt others, many mass shooters were also suicidal, with a plan to kill themselves or be killed during the event.

Another study, published in 2024, indicated that the percentage of shooters who died during shootings had doubled over the prior 20 years.

Czeisler, whose Ph.D. research investigated circadian rhythms and their disruption during the pandemic, wondered whether another biological rhythm — seasonality — might be important in exploring the question, allowing a broader view than prior studies more tightly focused on the shooters themselves.

In conducting the research, Czeisler used publicly available data from the Gun Violence Archive, the National Center for Health Statistics, and the Centers for Disease Control and Prevention, and analyzed 3,305 mass shootings and 360,827 suicide deaths between January 2014 and October 2021.

Mass shootings are significantly rarer than suicides in the U.S. and have a 613 percent difference between their summertime peak and winter valley. Suicides rose 12 percent between their valley and peak.

“Further research is really needed to look at each of these elements and additional potential elements that could account for why the seasonal patterns are so closely aligned,” Czeisler said. “But in the interim, we wanted raise awareness that these are temporally aligned and to advocate for both enhanced screening in the summer for suicidality and potential for gun violence and to ensure there are resources available from a healthcare perspective to respond to these extreme violence events.”

One potential follow-up study that might be illuminating can’t be done, Czeisler said. It would be instructive to look for similar patterns elsewhere, including nations in the Southern Hemisphere, to see whether annual patterns observed here are reversed, as would be expected if they are driven by seasonal variation.

The problem — though a good one to have — is that the large number of mass shootings seen in the U.S. is not replicated in other countries.

“It would be beneficial to have comparisons in other parts of the world where seasons are inverted, but fortunately, those data are not available because there aren’t enough mass shootings to study them,” Czeisler said.

If you are having thoughts of suicide, call or text 988 to reach the 988 Suicide and Crisis Lifeline.

One-third of young scientists who were planning careers in academic research changed their minds in the spring of 2025, according to a new survey, with fewer expecting to remain in higher education and in the U.S. to work or feeling that pursuing a Ph.D. was the right decision.

The survey, recently published as a working paper by the National Bureau of Economic Research, was taken at a time of tightening government policy on science funding and the presence of international students in the country. The government’s efforts in those areas have been stymied by the courts, at least initially.

But the new study suggests the high-profile campaign and lack of a final legal resolution have fueled uncertainty among younger scientists about the prospects for careers here. It also adds to researchers’ concerns about the disruption of the partnership between the U.S. government and universities that has driven the nation’s leadership in scientific discovery, innovation, and economic growth.

The survey, conducted in March 2025, showed that a third of the respondents who had previously said they were likely or very likely to pursue a career in academia no longer felt that way. Similarly, almost a fourth of those who had previously expected to stay and work in the U.S. no longer felt that was likely or very likely. And finally, when asked whether they were satisfied with having pursued a Ph.D., 17 percent fewer said they were either satisfied or neutral about the decision compared with six months earlier.

The survey occurred somewhat serendipitously.

A team including Raffaella Sadun, the Charles Edward Wilson Professor of Business Administration at Harvard Business School, and colleagues from the Massachusetts Institute of Technology and Monash University in Australia were in the midst of a large study of biomedical lab management practices when the Trump administration began its crackdown on government-funded science and higher education.

“We were running interviews with senior postdocs in the labs, and we started to realize we were in the middle of a pretty big policy change,” said Sadun. “The change was seeping into the interviews and, as we were talking to the postdocs, the anxiety was palpable. They started telling us how the uncertainty of the research funding environment, the policy changes, was affecting their own decisions.”

Sadun and colleagues had already compiled a sample of 2,466 biomedical research labs across the U.S. and, as strife between the administration and universities mounted, they understood they were in a position to analyze the extent of the disruption in real time.

So they reached out to more than 13,000 doctoral students and postdoctoral fellows who had taken part in the larger management study with a simple, six-question survey: “How likely are you to stay in academia?” “How likely are you to stay in the United States?” And “How satisfied are you with having pursued a Ph.D. in science?” They asked each question twice, once “for now” and again for, “six months ago.”

“We are running these interviews. We have this unusual access to a large number of labs and people. We had to run with this,” Sadun said. “It’s not perfect, but let’s try to measure as much as we can whether the current uncertain environment is having an impact on how they perceive or how they’re thinking about their future career moves.”

The authors acknowledged that the survey design — asking respondents how they would have answered a question six months earlier — may make it vulnerable to “recall bias,” in which respondents remember the past more favorably than they then viewed it.

The original management study, however, provided a check against that, since about a quarter of respondents had previously answered the Ph.D. satisfaction question before the end of 2024 — and the January 2025 start of the new administration. A comparison of those past responses with the recalled ones showed no sign of recall bias, according to the study.

The management researchers received responses from 916 scientists from 770 labs at 134 institutions.

The responses indicated that six months prior to the survey, 66 percent said they were likely to stay in academia, a figure that fell 22 percentage points, to 44 percent, by March 2025. Similarly, 93 percent of respondents said that six months earlier they had been likely or very likely to remain in the U.S. That number fell to 72 percent in March 2025. And while 92 percent of those who six months prior had been neutral or satisfied with their decision to pursue a Ph.D., that fell to 76 percent by March 2025.

The implications of the young scientists’ changed intentions are potentially important for U.S. science and more broadly, the nation’s economy, Sadun said.

She also acknowledged, however, that though she knows several scientists who have already relocated, broadly speaking, people don’t always follow through on their stated intentions. That’s something that can be studied in future research.  

“This is not great for research and it’s problematic from the perspective that R&D activities that are conducted in universities are an engine of growth,” Sadun said. “What you’re looking at is a significant shrinking of a population of scientists that you’ve created. They’ve spent time here, they’ve learned here. You’ve shaped this talent and this talent is telling you, ‘I want out.’”

This research was funded by the Alfred P. Sloan Foundation.

Dustin Tingley, a distinguished political scientist and leader in education innovation, has been named Harvard’s next vice provost for advances in learning, the University announced today.

“I am delighted that Dustin has agreed to serve as the next vice provost for advances in learning,” said University Provost John F. Manning. “He brings to this important role a wealth of experience in residential and online education, a deep understanding of technology’s potential to support academic excellence, and a strong commitment to innovation.”

As vice provost for advances in learning, Tingley will lead the University’s efforts in pedagogical innovation, digital education, and exploring the role of technology in supporting teaching, lifelong learning, and research on campus and around the world.

Tingley became interim vice provost in August 2025 and was deputy vice provost prior to that. He has long worked to develop, launch, and oversee initiatives to support teaching and learning at Harvard and beyond. Over the past decade, he also served as faculty director for the office’s Data Science and Technology Group, as well as for the Harvard Initiative for Learning and Teaching.

“It is an honor to continue to work alongside staff, faculty, and students in our collective pursuit of rigorous and engaging learning opportunities,” Tingley said. “The positive benefit we can bring to society through learning is a core mission of higher education institutions, and one where Harvard continues to have tremendous impact and opportunity.”

As a political scientist and public policy scholar with an interest in research methodology, Tingley’s research interests are broad, spanning international relations; the political economy of energy; data science and machine learning; and digital education. His recent work has included understanding how communities view changes in energy technologies, predicting students’ likelihood for online course completion, and applying machine learning to the social sciences.

Tingley has a joint appointment in the Government Department and the Harvard Kennedy School and teaches on the politics of climate and energy and international relations. In addition, he has taught on the business of the energy sector at Harvard Business School.

Outside of his research and teaching, Tingley has chaired Harvard’s Standing Committee on Climate Education and co-chaired the Faculty of Arts and Sciences Standing Committee on Public Service and Engaged Scholarship.

In 2012, he co-founded ABLConnect, an online database of active learning resources. He also previously served as director of graduate studies for the Government Department and faculty director of the Institute for Quantitative Social Science’s Undergraduate Research Scholar program.

Tingley holds a Ph.D. in politics from Princeton University and a Bachelor of Arts in political science with a minor in mathematics from the University of Rochester.

Early in Leah De Forest’s debut novel, New Yorkers pause to appreciate fat raindrops viewed from the comfortable side of a window. It’s miserable to be caught in the rain without an umbrella, but through the glass, it’s beautiful. 

“And aren’t so many things?” De Forest writes in “Any Kind of Known Tomorrow,” out Sept. 15. “Beautiful not just from a distance, but from the right distance.”

The novel takes place in the surreal early days of the COVID-19 pandemic, a time of makeshift facemasks, bleach-wiping groceries, and wondering just how bad things would get. It follows four women — a daughter, a mother, and a grandmother, along with an unexpected guest — sardined in a New York City apartment during the pressure-cooker weeks of the lockdown. The book joins a growing catalog of novels that tackle that bizarre period, including Ben Lerner’s “Transcription,” Ann Patchett’s “Tom Lake,” and Gary Shteyngart’s “Our Country Friends.”

De Forest, an instructor and academic adviser at Harvard Extension School, talks about the challenges and possibilities of revisiting the pandemic in fiction. The interview was edited for clarity and length.

Gazette: What made you want to engage with the very particular moment of the early pandemic?

Leah De Forest: I actually started writing this book in 2018. The draft I finished in 2019 was set in New York City, generally concerned with displacement and belonging, and it featured a character struggling with a novel disease.

In 2020, I started working on the next draft — and changed course to account for what was going on in the world. Everything felt unstable at that time, and I was interested in how people were dealing with that. Denial, anxiety, retreat: I’m sure we all saw lots of different responses. I took my characters and explored how each of them was metabolizing those early weeks.

Did you worry that readers wouldn’t be ready to, or wouldn’t want to, revisit such an intense and difficult time? 

When you’re writing a book, you’re writing the story that’s in front of you. This is what was in front of me. Later on, you might start thinking about whether there are readers for a book and what the marketplace is like.

Fiction has always been an important part of how I process the world. Literature has an ongoing role in representing a range of experiences — including, I think, the pandemic, which affected so many people. Any book can only get at a slice of any event or experience, right? My hope is that as people start to feel ready to read about the pandemic, they’ll find the right books for them.

What do you make of the subgenre of COVID-19 literature that’s come out in the last few years? 

Looking back, the pandemic can seem like a sort of bubble. We entered this strange time together, and then at some point — and exactly when was different for everybody, and for some people it’s of course still ongoing — we came out of it. In just beginning to talk to people about this book, I’ve noted that people often have a story to offer. It might be, “I went back home to be with my family for the first time in a decade and we fought all the time,” or “I had to quarantine in my childhood bedroom.” And of course, there was so much loss. There’s a lot to process.

In terms of books out there, I admired how Ann Patchett’s “Tom Lake,” for example, captured the sense of isolation people felt during the pandemic — and the way the family in that book has to retreat. The need for isolation folded the family back in on itself, in a way.

Your book follows three generations of women in the Parris family who each engage with the pandemic in different ways, so tell me a little bit about those women. 

Hadley Parris, the mother, is convinced at first that everything is going to be fine. Understandably, she thinks, “Well, surely this can’t be happening.”

Hadley’s mother, Aileen, gets kind of obsessed with the news. She tracks everything blow by blow — in some ways, she’s as fascinated by the performance as by the reality. As time goes on, though, she’s afraid. And Isabelle, one of the younger characters, is anxious about the future generally, so for her, it’s a confirmation of what she already knew: Things are bad. There’s tension between the characters, because you have one person saying, “Everything is going to be fine,” and somebody else saying, “Wait, I really don’t think this is going to be OK.”

The fourth primary character, Anna, is a teen girl whom the Parris family takes in in a moment of need. But Anna needs to keep working outside of the home to make ends meet, which Aileen perceives as a selfish exposure risk. Talk to me about that tension. 

I’m always interested in how structural factors, such as class, affect the way people interact. In Aileen’s case, she just doesn’t get other people’s perspectives, though she does sometimes try. That made her interesting to work with, because she’s often misunderstanding what’s going on and struggles to understand other people’s emotions. In this situation, although she might eventually come to understand that it’s not possible for someone to just quit their job, she’ll struggle to really feel why.

Through your omniscient narrator, you also dip into the heads of other people in this world, the man at the bus stop, a nurse at the hospital. Why did you choose to visit these other worldviews? 

The pandemic reminded us how connected we all are. People bonded with neighborhood groups and pods; it was, in many cases, a shared experience. But in other ways we were really isolated. The experience was not the same for everyone. In terms of the book, there was an interesting tension: These characters are in this very confined space, but the whole big world is always right out there, affecting them all very directly, in ways they can’t always see.

With every beat of the heart, every moment we’re alive, there’s so much going on. In our own bodies, of course, and also in our families, our communities, and beyond. Writing in this way allowed me to at least have the fantasy that I could access some of that complexity. An omniscient narrator can capture a sense of scope that isn’t always possible when you’re living in the moment. And I really enjoyed working in this mode. It’s the ideal fantasy, right? Like going to a dinner party and being able to read everybody’s minds at the table, what everybody’s thinking and not saying to the other person — as well as pop into the passing Toyota Corolla, see what’s going on in there, and maybe zoom into a distant warehouse, where someone’s packing broccoli or socks or bicycle parts.

The book and its characters are all wrestling with uncertainty in their own way. What do you hope readers take away about how we can relate to the discomfort of not knowing? 

So many things in this world are uncertain, and I think we all have to find our way to live with that. I don’t have a simple answer for how, which is probably why I wrote a book around it. But I’ll say this: The books I love the most show me something that I recognize in the world and maybe haven’t fully articulated for myself yet. I hope that readers who are interested in examining uncertainty, and the complicated relationships between mothers and daughters, and the strange and sometimes amazing ways we’re all connected, might find some of that in this book.