“Do you really think a technology that is so heavily used and understood by younger people, who are smarter and more flexible and more willing to learn new things, is going to harm those same young people forever? It just doesn’t add up.” — David Deming
As long as we’ve been collecting data, new college graduates had better odds of landing a job than the average American worker. But that flipped after the pandemic. Although the obvious suspect is AI, David Deming - labor economist and Dean of Harvard College - thinks the obvious suspect is probably innocent. In the first solo episode of The Context Window, he studies the scene of the crime: why the timing points to remote work rather than AI, how the pandemic quietly “containerized” white-collar work and made it legible to machines, and why he is willing to put a marker down that today is as bad as AI is ever going to be for young workers. The corporate ladder is good for developing talent among a chosen few. Now that ladder is crumbling, which may be a good thing in the long-run.
Listen on YouTube, Spotify, and Apple Podcasts
A programming note: this week I’m trying something different. The first episodes have been long biographical conversations, which I love, but a few of you wrote in to say you’d like to hear more about what I think. My instinct is to listen more and talk less - when I’m talking I’m not learning - but you asked, so we’re giving it a shot.
This one is a rant, affectionately titled “The Kids Are Alright.” As Dean I spend a lot of my time fielding worried questions from parents, alumni, and faculty: are the students okay? Short answer - yes! They’re doing so much better than the headlines suggest. The longer answer runs through the toughest job market for new graduates in a generation, a murder mystery about whether AI or remote work is to blame, and why I think the gloom about young people and AI is misplaced.
Please let me know what you think of the episode to david@thecontextwindow.com. Next week, we have Walter Isaacson on the Declaration of Independence coming up, just in time for the 250th anniversary of the nation.
[00:00] Cold open For the first time on record, a new college graduate is more likely to be out of work than the average American. Everyone blames AI. David thinks everyone is wrong.
[00:29] “The kids are alright” What David actually hears in the dining halls, why the students are doing better than the people who ask about them, and the one campus topic he’s saving for another day.
[02:44] Six dollars to one The bipartisan finding that the federal government spends roughly six dollars on retirees for every dollar on children, the broken bargain underneath the anxiety, and Samuel Moyn’s case for naming it.
[04:31] The employment picture A deliberately boring Fed chart, two lines that just crossed for the first time in thirty-five years, and what 5.6 percent really means.
[06:48] Is AI the culprit? Layoff narratives, the jump from 7 to 40 percent of job cuts blamed on AI in a single year, and why economists distrust what people say over what they do.
[08:58] The timing doesn’t fit David’s own survey work on how fast - and how quietly - people actually adopted AI, and why Claude Code, not ChatGPT, is the event a future economist will reach for.
[12:49] The better suspect: remote work The Lambert and Schindler “horse race,” and the stubborn fact that junior hiring started falling before ChatGPT existed, in four countries at once.
[15:16] Buy vs. build Why firms hire new graduates at all, the solar-panel logic of training someone up, and how the old Hollywood studio system explains what happens when you stop betting on unproven talent.
[22:57] Containerization How remote work wrote down the office’s unwritten rules, and why a metal box invented in 1956 is the key to what AI can and can’t automate.
[24:57] Where AI actually fits The marker David puts down - today is as bad as AI will ever be - and why the firms cutting juniors are eating their seed corn.
[27:10] AI-native firms, and the kids Flatter, more expert companies, the electricity-and-the-factory-floor analogy, and why David is betting on the young.
[29:46] Why a solo episode A note on the experiment: less of the guests this week, more of the Dean, and a standing invitation to tell him whether it works.
Data and trackers
The Labor Market for Recent College Graduates - the New York Fed’s quarterly tracker, David’s “one-stop shopping” for the recent-grad job market
Papers
Erik Brynjolfsson, Bharat Chandar and Ruyu Chen, “Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence” (Stanford Digital Economy Lab, 2025) - the ADP-payroll study using ChatGPT’s release as the event
Peter John Lambert and Yannick Schindler, “The Broken Ladder: AI, Remote Work, and Early-Career Hiring” (2026) - the horse race that finds remote work, not AI, explains the decline in junior hiring
Natalia Emanuel, Emma Harrington and Amanda Pallais, “The Power of Proximity to Coworkers“ - the Fortune 500 study showing seniors mentor juniors far more when they sit nearby
Alexander Bick, Adam Blandin and David Deming, “The Rapid Adoption of Generative AI“ - the first nationally representative survey of AI use
Aaron Chatterji et al., “How People Use ChatGPT“ (with David Deming) - on how much AI use is personal rather than work-related
F. Andrew Hanssen & Alexander Raskovich, “Does Vertical Integration Spur Investment? Casting Actors to Discover Stars during the Hollywood Studio Era” (2020)
Books
Samuel Moyn, Gerontocracy in America: How the Old Are Hoarding Power and Wealth - and What to Do About It (Farrar, Straus and Giroux, 2025)
Ideas and examples
Hyunjin Kim and Rembrand Koning, “AI-Native Firms” (Harvard Business School Working Paper, 2026) - smaller, flatter, more expert startups built around AI
Challenger, Gray & Christmas, May 2026 job-cut report - the source for the 7-to-40 percent jump in cuts attributed to AI
Malcolm McLean and the shipping container (1956) - the standardization that made global shipping automatable
United States v. Paramount Pictures and the unraveling of the Hollywood studio system - the build-vs-buy shift from developing unknowns to paying established stars
Further reading
David’s companion essay, “Remote work containerized the office, paving the way for AI”, on Forked Lightning
Credits
Host: David Deming, Danoff Dean of Harvard College
Executive Producer: Denise Koller Consulting Producers: Tim Smith and Jonathan Palumbo
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.