The first job is getting harder to land in America: What AI is doing to fresh graduates

For a long time, the first job in the US came with a basic reassurance: There would be a way in. It might not be a dream role. It might be repetitive, poorly paid, or only mildly related to what a graduate had studied. But it usually gave young people a starting point. That starting point now appears to be under pressure in parts of the labour market most exposed to AI. A recent Stanford study, based on monthly individual-level payroll records from ADP, the largest payroll software provider in the United States, tracked employment through September 2025 across millions of workers and tens of thousands of firms. It found that the strain is falling unevenly. In the most AI-exposed occupations, workers in the age bracket of 22 to 25 saw a 6% decline in employment from late 2022 to September 2025. Older workers in those same occupations, on the other hand, recorded 6% to 9% growth. Even after controlling for firm-level shocks, the researchers still find a 15 log-point decline in relative employment for young workers in the most exposed categories. The warning here is not that jobs are disappearing everywhere at once. It is that the market, in some corners, seems to be growing less patient with beginners.

Why fresh graduates are feeling the pinch first

New graduates usually get hired for the neat, manageable bits of white-collar work — the first draft, the basic analysis, the routine coding, the support task, the research clean-up. The Stanford study suggests this is exactly the zone where AI is becoming useful enough for employers to start rethinking headcount. Younger workers do more of this codified, repeatable work. Older workers, for all their flaws and inflated meeting calendars, are more likely to carry judgement, context, memory and the quiet practical sense that software still cannot fake very well. So when companies go looking for “efficiency”, they do not rush to cut the seasoned hand. They start lower down, shaving off the beginner layer. That is why the strain shows up first among workers aged 22 to 25. The problem here is not youth in itself. It is that the earliest tasks in a career — the very tasks through which people learn how work works — are becoming easier to automate, easier to redistribute, and, from an employer’s point of view, easier not to hire for at all.

The story changes when AI works as a tool

This is where the story becomes more nuanced than the usual alarm about AI taking jobs. The Stanford paper does not suggest that every occupation touched by AI is turning hostile to young workers. Instead, it hints at something more precise: The effect depends on whether AI is being used to replace work or to support it. Where it is used mainly to automate tasks, early-career employment weakens. But in fields it is used more to augment human tasks, the pattern is less grim. The study finds that the occupations with the highest estimated augmentation share were among those with the fastest employment growth for young workers. It also shows a broader split: Close to 70% of occupations in the lowest AI-exposure group recorded rising early-career employment between October 2022 and September 2025, compared with less than half in the highest-exposure group.For fresh graduates, that distinction is not semantic. It goes to the heart of whether or not a role still exists as an entry point. If AI helps a junior analyst work faster, a new coder test more efficiently, or a fresher handle more volume with supervision, the job may survive and even improve. But when AI starts taking over the simpler, lower-risk work, the reason to hire a beginner begins to fade. The question for an employer in such a case is no longer how to train a beginner, but whether the beginner is needed at all.That is the more disquieting implication here. Entry-level roles have never been valued only for immediate output. They also serve as training grounds, the place where competence is built slowly and sometimes inefficiently. AI, with its capacity to substitute efficiently, unsettles that bargain. It tempts firms to treat junior jobs less as an investment for future capability and more as a cost that can be trimmed.

In AI-exposed jobs, a weak economy hits beginners first

A weak economy usually hits beginners first. That is hardly a revelation. When companies get uneasy, they do not always rush into layoffs. They often begin more quietly than that: A smaller trainee batch, a pause on junior hiring, a decision to stretch the existing team a little further. At the onset, it all looks like a logical business decision. The work is still getting done but what goes missing is the opening someone was supposed to walk through.That is what makes the Stanford finding worth pausing over. The researchers are asking a simple question: Are young workers doing worse merely because the market has cooled, or is something more specific happening in AI-exposed jobs? So they compare workers within the same firms over time, instead of blaming every dip on some vague economic mood. Even after that, workers aged 22 to 25 in the most AI-exposed occupations still show a 15 log-point decline in relative employment compared with those in the least exposed ones. For older workers, the pattern is far weaker.That is the real discomfort here. This is not just a weak economy doing what weak economies do. It is a weak economy becoming more choosy about whom it shuts out first. And in AI-exposed work, that seems to be the beginner. The technology does not out entire professions, it only makes employers feel they can manage with one less newcomer. Once that instinct settles in, the slowdown begins to decide who gets the first chance at all.

The paycheque holds, the opening shrinks

For fresh graduates, the first warning sign is not a weak salary offer. The Stanford paper, in fact, notes little difference in annual base compensation trends across age groups and levels of AI exposure. The trouble is showing up earlier: In the job that is quietly not posted, in the junior role that stays vacant a little longer or in the beginner’s work that gets absorbed, stretched, or handed to software before it turns into an actual opening. Those already inside may not notice much at first. For fresh graduates, however, the shift is immediate.

What starts as hiring caution can end as a talent shortage

The US market for recent graduates was already turning less forgiving. Federal Reserve Bank of New York data shows unemployment among recent college graduates rising from 4.0% in Q4 2022 to 5.7% in Q4 2025. What the Stanford study adds is a sharper point: In occupations most exposed to AI, the axe is falling more heavily on the youngest workers. This suggests that the market is cooling selectively. The real damage in this scenario is hardly visible. It begins to show when the labour market starts eating its own future. As the beginner roles shrink, the loss is not confined to one unlucky graduating batch. It hits the system’s ability to produce skilled workers for the future. The first job is more than an income source. It is where graduates turn into professionals through routine, supervision, correction and time. If that layer of work is increasingly automated, redistributed or withheld, firms may save money in the short term. But they also narrow the pipeline from which the next generation of experienced workers emerges. The result is a market that becomes more exclusionary at the bottom and more anxious about talent higher up. Here lies the contradiction: Employers lament skill shortages even as they help dismantle the place where skills are first built.