Key facts
- Employment for workers aged 22-25 in the most AI-exposed occupations is 19% below peers in less exposed fields.
- This gap has widened from 13% in the previous year's study.
- The phenomenon is driven by lower hiring rates for entry-level workers in AI-impacted fields.
- Jobs where AI is used for 'automation' show the worst relative employment levels for young workers.
- Occupations with higher codified knowledge have slower entry-level employment growth.
- Jobs with a higher share of college graduates show more muted differences between AI-exposed and less-exposed occupations.
New research from Stanford University economists suggests that artificial intelligence is significantly impacting entry-level jobs, particularly for younger workers. The updated study, "Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence," released in August 2026, indicates that employment levels for workers aged 22 to 25 in the most AI-exposed occupations have fallen by 19 percent compared to their peers in less AI-disrupted fields. This gap has widened from 13 percent a year prior.
The researchers utilized anonymized payroll data from ADP and assessed AI exposure using previous impact gauges and Anthropic's Economic Index, which analyzes the use of AI models like Claude. While overall employment across the economy shows minimal differences between AI-exposed and less-exposed jobs, the disparity is stark for young workers. Since 2022, employment in the top 40 percent of AI-impacted jobs for this age group has decreased by approximately 11 percent, while employment in the least AI-impacted jobs grew by 10 percent.
This trend is primarily attributed to lower hiring rates for entry-level positions in AI-impacted fields, rather than increased layoffs or employee departures. The study also found that jobs where AI is used for 'automation' are experiencing the most severe relative employment declines for young workers. Conversely, jobs that use AI for 'augmentation' show more mixed results, and occupations with higher codified knowledge are seeing slower entry-level employment growth.
Furthermore, the research suggests that higher education may act as a buffer against AI's employment effects. Occupations with a greater proportion of college graduates exhibit less pronounced differences in employment between AI-exposed and less-exposed sectors. Lead researcher Erik Brynjolfsson expressed concern over these trends, warning of a future where existing jobs persist but the entry points for new careers may effectively close.