If A.I. Takes the Jobs of Young People, How Will They React?
In the first column in this series, I wrote about how the economic reality for young people in this country doesn’t exactly match the bleak picture that is sometimes painted of it. In the second, I offered a theory about the discontent that is nonetheless fuelling a new populist economic politics among college-educated Gen Z-ers and millennials. In short, these voters believe, not without reason, that the old-fashioned trappings of middle-class life—homeownership, stability, a sense of purpose—are out of reach. The frustration and disappointment of the young, educated, and downwardly mobile middle class has become the defining force in Democratic politics.
There’s a big question hovering in the background of all this, which I have not yet addressed, but which is certainly on the minds of many of these voters. That is, will A.I. make everything worse for them? More specifically, is A.I. going to make it even harder for young people to find jobs, thus adding to their dissatisfaction?
According to new findings from the Federal Reserve Bank of New York, the unemployment rate for recent college graduates between the ages of twenty-two and twenty-seven is 5.7 per cent, which is higher than the rate for workers at large. This is a relatively new phenomenon—for four decades straight, prior to 2022, young college graduates had lower unemployment numbers than the general population did. When you compare the unemployment rate of young college graduates with that of college graduates in general, the gap is even larger, and, during the past three years, it has grown, albeit by a relatively small degree. Given the strength of the over-all labor market, some observers have theorized that companies in white-collar industries—which, so far, tend to have the most exposure to A.I.—might be leaving some entry-level positions unfilled while they determine whether robots can replace, say, a paralegal or a junior analyst or a cub copywriter.
The fact that the young-college-graduate unemployment rate keeps detaching itself from other employment trends may not be the result of A.I. For one thing, this began happening well before the introduction of ChatGPT: since 2012, the recent-college-graduate unemployment rate has stayed relatively stagnant, while the rate for pretty much everyone else has improved. For another, the pandemic-era shift to remote work almost certainly convinced companies that some positions needn’t be filled at all. This probably explains more about the latest numbers than the offloading of jobs to A.I. does—although the arrival of A.I. may have played a part in those decisions.
It is hard, so far, to find conclusive evidence that A.I. is already replacing young college workers. At the same time, it seems ridiculous, or at least a bit irresponsible, not to acknowledge that many industries probably could replace a healthy portion of their workers with A.I., or that jobs such as software engineering have been irrevocably changed in the past two years. Focussing simply on tech, it seems impossible to believe that there will be no labor impact, given that many programmers now simply tell Claude or Gemini to do some stuff in Python rather than doing it themselves. Last year, a Stanford Digital Economy Lab study titled “Canaries in the Coal Mine” showed that the number of employed twenty-two- to twenty-five-year-old software developers fell by roughly twenty per cent between 2022 and 2025. The study also found that occupations with more exposure to A.I. experienced a greater decline in the employment of twenty-two- to twenty-five-year-olds than those with less exposure. And the authors found that much of this job loss was concentrated in fields that used generative A.I. for automating, rather than augmenting, some tasks.
If you accept these findings, and believe that they are quiet indicators of a coming reckoning, then the question is whether what has happened to companies with a lot of exposure to A.I. will spread to other industries. Will it reach management consulting or payroll management or insurance or architecture or media?
The answer, I imagine, is yes, at least for a while. Not necessarily because A.I. actually can do most white-collar entry-level jobs well, but because companies are incentivized to cut payroll—and because management at these companies tends to live and die by their handling of new technology, which makes them uniquely subservient to the dictates of Silicon Valley. This might not mean mass layoffs across every white-collar industry, obviously. But it could mean that an advertising agency that once had, say, forty entry-level positions might soon have twenty-five, ten of which might be, effectively, low-paid internships.
The youth resistance to this coming reality has, in any case, already begun. Commencement speakers who mention A.I., or who, God forbid, work for A.I. companies, are getting roundly booed by students. More telling, many polls show that college students and recent graduates are more skeptical of A.I. than the general public is. A Gallup poll surveying Gen Z’s emotions regarding A.I. found that “anger” had increased over the course of the year, while “excitement” had dropped noticeably. Nearly half of the employed respondents said that A.I.’s risks for the workforce “outweigh the potential benefits.” An analysis by Pew found that forty-eight per cent of eighteen- to twenty-nine-year-olds were “skeptical about AI’s future impact,” compared with just thirty-five per cent of those above the age of sixty-five.
These attitudes have already begun to affect political results. In the Wisconsin governor’s race, for instance, much of the support for the presumptive Democratic nominee, Francesca Hong, stemmed from her maverick stance against data centers. Hong not only supports a conditional moratorium on building data centers in Wisconsin but has also proposed eliminating automatic tax exemptions from A.I. data centers that are attempting to operate in the state. Unsurprisingly, Hong polls strongly among young voters. It’s possible that data centers are becoming an electoral dividing line—young voters, in particular, might begin to oppose any candidate who they suspect has been bought off by big tech.
Like some of the other trends I’ve discussed in this series, the brewing youth revolt against A.I. is still a bit formless. Young people are not exactly immiserated; nor have all their jobs been taken over by Claude, at least not yet. But A.I. automation is another reason that many young people may believe the future is hopeless and that an American renewal will not take place in their lifetimes.
For now, this fuzzy discontent has led to a relatively soft populism, peddled by candidates who are promising universal health care and the cancellation of debt, for instance. It hasn’t given rise to the type of rhetoric that defined, say, the rabble-rousing governor and senator Huey Long, a century ago. As candidates get close to election time, even those furthest on the left tend to cut back on the anti-billionaire talk and focus on affordability or, more recently, on Gaza and aid to Israel. Bernie Sanders never stops campaigning on the same anti-oligarch message, but we still haven’t seen an insurgent populist candidate who sounds truly revolutionary, or, perhaps more pressing, who talks with real fury about the end of middle-class hope in America. To date, the rhetoric of American carnage has been relegated to the right, with its promise to “make America great again.”
But the anger over data centers and the growing anxiety about a technology that will take away young people’s jobs may be what pushes a truly populist politics into the liberal mainstream. If unemployment rises among young people, and the technocratic promises of the past are even further diluted, the fairly mild forms of collectivist politics that now exist on the left might start sounding more, in their way, like the America First populism you hear on the right. Consider that new college graduates, with an unemployment rate a little above five per cent, have already helped notch electoral wins for left-wing candidates, not only in New York City but also in Colorado and Michigan—and Wisconsin may soon follow. What will happen if that number hits ten per cent, and if there’s more robust evidence that the emerging technological powers in this country have decided that employees between the ages of twenty-two and thirty are no longer in their plans? ♦