Artificial intelligence is often discussed as a threat to experienced professionals whose work involves writing, programming, or analyzing information. But some of the earliest signs of disruption in the US labor market are appearing among people who have barely started their careers.
Young workers are finding it hard to get hired in occupations where AI can take over a growing share of everyday tasks. Research from Stanford University’s Digital Economy Lab suggests the problem is already measurable.
Employment among workers aged 22 to 25 in occupations highly exposed to AI is falling behind that of their less-exposed peers. The researchers have not found evidence of widespread economy-wide job displacement, although young workers in AI-exposed occupations are experiencing a significant employment gap.
Young workers are losing ground before mass layoffs begin
Stanford’s researchers analyzed ADP payroll records covering millions of US workers through June 2026. Their August update found that employment for 22- to 25-year-olds in AI-exposed occupation groups was about 19% below the level it would have reached by keeping pace with similarly aged workers in less-exposed fields.
This comparison does not mean one in five young workers has lost a job. The gap has widened since the researchers first reported it in 2025. The study points mainly to fewer young people being hired, not mass dismissals. Experienced workers in the same occupations have not shown a comparable shortfall.
Entry-level jobs also provide opportunities to learn workplace practices and build experience. This pattern also varies with how companies use AI. Employment has weakened most in occupations where the technology automates tasks previously done by people. In jobs where AI is used to assist workers, employment has been flat or has grown, particularly for experienced employees.
The damage is concentrated in jobs AI can perform, not every AI-exposed occupation
Software development and customer service stand out in Stanford’s October update. Employment among the selected age group in these fields has fallen by more than 3% annually since ChatGPT’s November 2022 release.
The decline recently reached 4.4%, according to the university’s dashboard. These figures concern young workers in two fields, not everyone employed there. A Census Bureau working paper offers another indication of pressure on new entrants.
Researchers compared graduates from majors with different levels of AI exposure. Among graduates from the most exposed tenth of majors, the likelihood of initial employment fell by five percentage points, while initial full-quarter earnings were 13% lower.
The earnings differences reflected weaker pay within industries and a shift toward lower-paying sectors. A separate Census analysis found that regression-adjusted employment among workers ages 22 to 24 in the most AI-exposed quintile of industry-state groups fell 12% over 10 quarters, although some divergence predated ChatGPT.
AI exposure measures how much of an occupation’s work could be affected, not whether an entire job can be handed to a chatbot. Many roles require judgment and collaboration.
The evidence points to a warning sign, not an economy-wide jobs crisis
Other research cautions against attributing every change to AI. A Federal Reserve Bank of New York analysis of job postings found little evidence of a distinct AI-related fall in demand. Vacancies in more exposed occupations were already declining before ChatGPT appeared, while postings for junior and senior roles moved in parallel.
A September working paper from the National Bureau of Economic Research also found no statistically significant increase in relative unemployment among recent college graduates during summer 2026, compared with older graduates or young people without degrees.
The studies measure different things. One study tracks employment within occupations, while the other examines unemployment among recent graduates.
The warning is not that AI has eliminated jobs across the economy, but the first rung of the career ladder may be harder to reach where software handles routine work. If employers automate tasks once assigned to junior staff, opportunities to gain experience could shrink.
Sources: United States Census Bureau, Stanford University Digital Economy Lab, National Bureau of Economic Research
Affiliate links on Cult of Mac may earn us and our partners a commission.