In a moment with many important economic questions and fears, I continue to find this among the more interesting mysteries about the US economy in the long run: Is artificial intelligence already taking jobs from young people?
If you’ve been casually following the debate over AI and its effect on young graduates’ employment, you could be excused for thinking that the answer to that question is “possibly,” or “definitely yes,” or “almost certainly no.” Confusing! Let’s review:
Possibly! In April, I published an essay in The Atlantic that raised the possibility that weak hiring among young college graduates might indicate an AI disruption. My observation started with an objective fact: The New York Federal Reserve found that work opportunities for recent college graduates had “deteriorated noticeably” in the previous few months. Among several explanations, including tight monetary policy and general Trumpy chaos, I considered the explanation that companies might be using ChatGPT to do the work they’d historically relied on from young college grads. As David Deming, an economist and the dean of undergraduate studies at Harvard University, told me: “When you think from first principles about what generative AI can do, and what jobs it can replace, it’s the kind of things that young college grads have done” in white-collar firms. Definitely yes! Soon after my essay went up, several other major news organizations and AI luminaries endorsed even stronger versions of my hedged claim. The New York Times said that for some recent graduates “the A.I. job apocalypse may already be here.” Axios reported that “AI is keeping recent college grads out of work.” In a much-discussed interview predicting a labor “bloodbath,” Anthropic CEO Dario Amodei made the audacious forecast that AI could wipe out half of all entry-level white-collar jobs within the next five years. By June, the narrative that AI was on the verge of obliterating the college-grad workforce was in full bloom. Until … Almost certainly no!: As AI panic reached its fever pitch, several whip-smart analysts called the whole premise into question. A report from the Economic Innovation Group took several cuts of government data and found “little evidence of AI’s impact on unemployment,” and even less evidence that “AI-exposed workers [were] retreating to occupations with less exposure.” In fact, they pointed out that “the vast majority of firms report that AI had no net impact on their employment.” John Burn-Murdoch at the Financial Times pointed out that “the much-discussed contraction in entry-level tech hiring appears to have reversed in recent months.” The economic commentator Noah Smith synthesized even more research on this question to reach the conclusion that “the preponderance of evidence seems to be very strongly against the notion that AI is killing jobs for new college graduates, or for tech workers, or for…well, anyone, really.”
To be honest with you, I considered this debate well and truly settled. No, I’d come to think, AI is probably not wrecking employment for young people. But now, I’m thinking about changing my mind again.
Last week, I got an email from Stanford University alerting me to yet another crack at this question. In a new paper, several Stanford economists studied payroll data from the private company ADP, which covers millions of workers, through mid-2025. They found that young workers aged 22–25 in “highly AI-exposed” jobs, such as software developers and customer service agents, experienced a 13 percent decline in employment since the advent of ChatGPT. Notably, the economists found that older workers and less-exposed jobs, such as home health aides, saw steady or rising employment. “There’s a clear, evident change when you specifically look at young workers who are highly exposed to AI,” Stanford economist Erik Brynjolfsson, who wrote the paper with Bharat Chandar and Ruyu Chen, told the Wall Street Journal.
In five months, the question of “Is AI reducing work for young Americans?” has its fourth answer: from possibly, to definitely, to almost certainly no, to plausibly yes. You might find this back-and-forth annoying. I think it’s fantastic. This is a model for what I want from public commentary on social and economic trends: Smart, quantitatively rich, and good-faith debate of issues of seismic consequence to American society.
To more deeply understand the new Stanford paper, I reached out and scheduled an interview with two co-authors, Erik Brynjolfsson and Bharat Chandar. A condensed and edited version of our interview is below, along with careful analysis of the most important graphs.
Thompson: What’s the most important thing this paper is trying to do, and what’s the most important thing it finds?
Erik Brynjolfsson: There has been a lot of debate out there about AI and jobs for young people. I was hearing companies telling me one thing while studies were telling me another. I honestly didn't know the answer. We went at this with no agenda.
When we were able to slice the data, lo and behold, subcategories of high-exposed jobs like software developers and customer service agents for people aged 22 to 25 saw a very striking decline in employment in the last few years.
... continue reading