Artificial intelligence may be making it increasingly difficult for young people to get their start in some professions, according to new research from Stanford University.
Researchers analysed payroll data covering millions of workers and found that employment among people aged 22 to 25 in occupations highly exposed to AI is now about 19% lower than it would have been if their employment had kept pace with similarly aged workers in less AI-exposed occupations.
Growing gap
The gap has continued to widen. In the previous version of the research, based on data through July 2025, the shortfall was estimated at 15%. The latest analysis extends the data through June 2026.
The researchers found little evidence of a comparable employment decline among more experienced workers in the same occupations. Instead, the change appears to be happening largely through hiring: companies are bringing in fewer young workers rather than laying off large numbers of existing employees.
The pattern is particularly pronounced in jobs where AI can automate tasks performed by people. By contrast, employment has remained stable or increased in occupations where AI is more likely to complement workers rather than replace parts of their work.
The findings could have implications beyond the immediate job market. Entry-level positions have traditionally allowed workers to gain the experience and practical knowledge needed to progress into more senior roles. If those opportunities become less common, young workers could find it harder to build the experience needed for future careers.
Codified knowledge
The researchers also found a difference between jobs based largely on codified knowledge — information that can be written down, taught and reproduced — and those that depend more heavily on tacit knowledge gained through experience and mentorship. Employment among young workers has fallen more in occupations relying heavily on codified knowledge, while experienced workers have fared better in roles where practical knowledge is more important.
However, Stanford researchers stress that the findings do not prove AI is causing the decline.
The study identifies a strong and growing association between AI exposure and employment among young workers, but other factors affecting the labour market could also be contributing. The researchers say the employment gaps become smaller when differences in education are taken into account, and some of the trends were visible before generative AI became widespread.
For now, the evidence does not point to an economy-wide wave of AI-driven job losses. Instead, it suggests that the effects may be appearing first at the bottom of the career ladder, where young workers are competing for roles that increasingly involve tasks AI can perform.
