Young adults entering the workforce are bearing the largest effects associated with artificial intelligence, according to updated research from a Stanford University team. The August 2026 edition of the paper, titled “Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence,” updates a paper published last year with fresh data and refined statistics. The researchers found that the employment trends previously identified for entry-level workers are persisting and expanding. Employment levels for workers aged 22 to 25 in the most AI-exposed occupations are now 19 percent below those of their peers in fields less exposed to AI disruption. To calculate these figures, the Stanford team analyzed a large subsample of anonymized, high-frequency payroll data regularly aggregated by the human-resources management company ADP. The team then rated each occupation’s exposure to AI disruption using two measures. The first was a potential labor-market impact gauge established by earlier researchers. The second was the Anthropic Economic Index, which examines how occupations actually use the Claude model in everyday work. Google released a similar report last month based on occupational use of Gemini. Across the full economy, the researchers found little to no difference in relative overall employment between the jobs rated most and least affected by AI. The picture changed when they separated workers aged 22 to 25. Since 2022, employment in the top 40 percent of AI-impacted jobs for that age group has fallen by about 11 percent. By contrast, total employment for those young workers grew by 10 percent in the 60 percent of jobs with the least AI impact over the same period. A closer examination of the payroll data found that the decline is mainly appearing through lower hiring rates for entry-level workers in AI-impacted fields, rather than through more dismissals or workers leaving their jobs. The researchers also found that the effects for this age group were seen mainly in lower overall employment, rather than lower pay rates for people who obtained work. Not every form of AI exposure affects jobs in the same way, the study found. The Anthropic Economic Index differentiates between “automative” queries, in which AI fully replaces work previously done by a person, and “augmentative” queries, in which AI helps workers with tasks for which they are still needed. Accountants, auditors, receptionists and information clerks were among the jobs judged most susceptible to automation. Chief executives and registered nurses were among those most often using AI as an aid to their work. Occupations where automation-oriented AI use is prevalent are showing the worst relative employment levels for entry-level workers. The researchers also suggested that entry-level workers may be especially exposed in jobs requiring heavily “codified” knowledge: formal, standardized and documented knowledge that can be taught through education, textbooks or written procedures. That contrasts with “tacit” knowledge, which is acquired through practice, mentorship and repeated exposure to real situations. The data found slower entry-level employment growth in occupations with more codified knowledge, while occupations with more tacit knowledge had faster employment growth for mid-career and senior workers. To test that theory, the team used the formal education required in the O*NET occupational database as a proxy for how much an occupation relies on codified knowledge. Higher education may still reduce the employment effects identified in the study. Occupations with a higher share of college graduates showed smaller differences between more-exposed and less-exposed occupations. In fields with few college graduates, the least AI-exposed occupations grew while the most exposed occupations declined in employment. In a recent interview with The Washington Post, lead researcher Erik Brynjolfsson warned that the trends suggest a near future in which jobs held by people employed before the AI era largely remain, while many jobs for the incoming working-age cohort begin to disappear. He said the entry-level effects measured by the team are real, persistent and widening, and expressed concern about a labour market that keeps its overall employment level while quietly closing the route into a career for people starting out. For Somali graduates, young professionals and members of the Somali diaspora seeking work online, the findings point to a tougher first step into occupations exposed to AI. Training that combines formal study with practical experience, mentorship and work that still requires human judgement may matter as entry-level opportunities shift.