Young People Are Falling Behind, but Not Because of AI
· The Atlantic
The job market for young people is brutal. Is AI to blame?
Visit newsbetting.club for more information.
ChatGPT was released in late 2022. Since then, the unemployment rate for recent college graduates has shot up to near 6 percent, its highest level in more than a decade, setting aside the 2020 pandemic spike. That’s true even though the overall unemployment rate is about 4 percent.
Many observers have interpreted these data as evidence that AI has begun displacing entry-level white-collar professionals. “AI Is Wrecking an Already Fragile Job Market for College Graduates,” reads a representative Wall Street Journal headline from last year. But the AI-stealing-the-jobs story turns out to be wrong—or at least premature. Almost every data point available suggests that whatever is holding back young people in the job market has nothing to do with AI. Whether that’s good or bad news is a much tougher call.
The case that AI is already stealing young people’s jobs is based on a statistical mirage. Historically, recent college graduates have had a much lower unemployment rate than the average worker. Since ChatGPT was released, however, the unemployment rate for this group has risen nearly twice as fast as the overall number. Because AI is best suited to replacing white-collar workers, this trend is what you’d expect to see if AI was having a labor-market impact.
[From the March 2026 issue: America isn’t ready for what AI will do to jobs]
But the unemployment rate can be a highly misleading statistic. The Bureau of Labor Statistics counts individuals as “unemployed” only if they’ve actively looked for work in the past four weeks; otherwise, they are removed from the data set. The rationale is to avoid counting people who don’t actually want a job, such as students, retirees, and stay-at-home parents. But it also excludes people who want to work but have stopped looking for a job.
The economists Adam Ozimek and Nathan Goldschlag recently took a deeper look at the data and found that a significant number of young workers without college degrees had simply given up looking for a job, artificially improving the unemployment rate for young workers without a degree and thereby giving the appearance that recent college graduates were doing uniquely poorly. This is the labor-market equivalent of a school’s worst-performing students simply not showing up on standardized-test day.
Using a different employment measure that includes all working-age adults 25 and younger (except full-time students), Ozimek and Goldschlag found that those without degrees have experienced an even worse decline than their college-educated peers since 2023. “It turns out the labor market for young people—all young people—is even worse than we thought,” Goldschlag told me. “That makes me doubt that this is an AI story.”
The fact that recent college graduates are doing better than young people without a degree doesn’t quite rule out the AI labor-market theory, because the gap between the two groups—the advantage that recent grads enjoy—is much smaller than it used to be. But the recent-graduate premium peaked during the Great Recession and has been falling steadily ever since, well before the introduction of ChatGPT. This timing suggests that the shrinking new-grad gap is a much longer-term story about supply and demand. The percentage of young people with a bachelor’s degree has risen by about a third since 2008, and most of that increase has stemmed from expanded enrollment at less selective universities. As David Deming, an economist at Harvard, pointed out to me, this means that the number of graduates competing for jobs has increased at the same time that the skill level of the average graduate has fallen. “This is the same thing that happened with high school 50, 60 years ago,” Deming said. “A high-school diploma used to confer huge advantages, but then it became so ubiquitous that the advantages went away.”
Meanwhile, research from the San Francisco Federal Reserve has found that the share of online job postings seeking workers with a college degree has declined since 2010—possibly because after that point, digital technologies were so widespread that having a college degree was no longer a prerequisite for using them.
Perhaps the most compelling reason to doubt that AI is to blame for the job market is that workers at the highest risk of AI displacement aren’t seeing the worst outcome. In an August report, Goldschlag and his co-author, Sarah Eckhardt, evaluated five different measurements of which occupations were most exposed to AI-related disruption to see whether any of them correlated with changes to employment outcomes from 2022 to 2025. “No matter how we cut the data,” they concluded, they didn’t “see any meaningful AI impacts in the labor market.” The economist Ernie Tedeschi has shown that since June 2023, unemployment for young workers has increased the most for those in occupations least exposed to AI, such as construction workers and fitness trainers. Most other studies have come to similar conclusions. “I’m very open to the possibility that AI could displace entry-level workers,” Martha Gimble, the executive director of the Yale Budget Lab and a co-author of one of these analyses, told me. “But we’re just not seeing it show up anywhere in the data.”
In fact, AI might be increasing employment for college-educated workers. One 2025 analysis found that since the release of ChatGPT, recent graduates in sectors with more AI usage have experienced slightly better employment outcomes than they did prior to its release. In an August survey of businesses from the Federal Reserve Bank of New York, a larger share of firms reported hiring more workers due to AI than reported hiring fewer. Even software developers—universally considered the canary in the coal mine for the AI jobs-pocalypse—are currently employed at their highest levels ever. “We use AI in all kinds of ways—coding, engineering, you name it,” Tedeschi, who serves as chief economist at the payment-processing company Stripe, told me. “And that hasn’t stopped us from bringing on new young people. If anything, our hiring has accelerated because of it.”
If AI isn’t to blame for the terrible job market for young people, then what is? In nearly every sector of the economy, the pace of hiring has slowed to levels last seen shortly after the Great Recession. A job market with few hiring opportunities is especially punishing for young people entering the workforce, including those with a college degree.
[Annie Lowrey: The job market is hell]
The slowdown, which I’ve previously referred to as “the big freeze,” began to emerge in mid-2022. The initial explanation was that employers, traumatized by the frenetic job-switching of the pandemic years, had decided to hold tightly on to their existing workers, preventing new positions from opening up. But the fact that the big freeze has persisted for as long as it has points to a second explanation: uncertainty. The fear of an impending recession and trepidation about the outcome of the 2024 election caused companies to pause plans to make new investments, open new locations, or launch new products—all of which meant less need to hire new employees. “Hiring young people doesn’t bring a lot of benefits right away—it is really an investment in the future,” Deming told me. “So if you’re unsure about what that future will look like, then that’s one of the first things you stop doing.”
When I spoke with hiring managers and employers a little more than a year ago, they were generally hopeful that, with the election decided and inflation defeated, the broader economic uncertainty would finally ease. Instead, they got an ever-changing set of tariffs and trade deals, attacks on the Federal Reserve and statistical agencies, higher-than-expected inflation, and, most recently, a foreign conflict threatening to upend the global oil market. Since Donald Trump took office, the Economic Policy Uncertainty Index—the most widely cited measure of economic-policy uncertainty—has reached its highest sustained levels ever. “You hear the same basic story from basically every employer, every hiring manager, right now,” Guy Berger, a senior fellow at the Burning Glass Institute, told me. “It’s impossible to predict where the economy will be in a few months, let alone a year or two. So we might as well just wait and see.” (Several economists, including Berger, told me that the looming question of AI’s labor-market impact could be part of this greater constellation of uncertainty, but that’s hard to see concretely in the data.)
In recent months, AI companies have been unveiling new agents capable of performing a wide range of tasks currently performed by entry-level workers. In that sense, the fact that the young-adult labor market is already weak isn’t exactly comforting. But if there’s a more hopeful takeaway from the data, it is that developments in the labor market are complicated and can’t be easily predicted. The experts I spoke with still can’t say for certain why young people have struggled so much in recent years; they have only theories. The mysteriousness of the labor market is itself concerning, but should also give us pause before we make big, scary predictions about AI with any certainty. No one even knows for sure what is happening right now, let alone what will happen in the future.