The employment gap between young workers in AI-exposed jobs and their peers in less-exposed roles has grown to 19%, according to a new update published this month by the Stanford Digital Economy Lab, a finding that points to a measurable structural shift in who is getting hired in the early stages of a career.
What the data show
Young workers in AI-exposed occupations are increasingly falling behind their less-exposed peers. Employment among workers ages 22, 25 in highly AI-exposed occupations now stands about 19% below where it would be if it had kept pace with employment among similarly aged workers in less-exposed occupations.
This divergence has widened steadily since researchers first documented it in August 2025: by this same measure, the shortfall was 15% at the July 2025 data vintage and is 19% as of June 2026.
The adjustment appears to operate primarily through reduced hiring of young workers rather than increased separations. Companies are not laying off entry-level workers in large numbers so much as simply not bringing them on in the first place.
Researchers do not see widespread, economy-wide job displacement associated with AI. The gap is concentrated at the entry-level end of the age spectrum, and experienced workers show no comparable shortfall.
Where the declines are concentrated
The declines are concentrated in occupations where AI usage tends to automate human tasks. In occupations where AI is used more to complement workers, employment is flat or rising, particularly among more experienced workers. That combination is not naturally predicted by explanations based solely on interest rates, education, or remote work.
In absolute terms, employment of workers ages 22 to 25 in the most AI-exposed occupation quintiles declined between November 2022 and June 2026, while employment of the same age group in the least-exposed quintiles grew over the same period.
When researchers examined the data by gender, there is some evidence that the employment growth gap between men and women is slightly higher for more exposed occupations, though differences across quintiles are modest and appear to be driven primarily by occupational composition rather than disparate trends within given sets of occupations.
The findings draw on a revised version of “Canaries in the Coal Mine? Six Facts About the Recent Employment Effects of Artificial Intelligence,” using payroll data from ADP. The collaboration between ADP Research and the Stanford Digital Economy Lab tracks employment across occupations and worker groups with different levels of AI exposure. Earlier analysis found that, when including the broadest set of controls, declines in AI-exposed occupations become statistically significant in 2024.
Why entry-level workers bear more of the burden
The pattern fits a broader theory about how automation tends to spread through labor markets. Junior workers typically handle the most routine, task-specific portions of a job, precisely the functions that generative AI tools automate most readily.
Experienced employees, by contrast, bring judgment, client relationships, and institutional knowledge that current AI systems replicate with far more difficulty.
The employment gap for young workers in highly AI-exposed occupations has continued to widen through mid-2026. The paper uses November 2022 as its employment baseline, meaning the subsequent decline extends well below that reference point rather than merely reversing a pandemic-era distortion.
The result lands on top of a broader period of labor-market caution. Nonfarm payroll employment fell by 23,000 in July 2026, with the unemployment rate at 4.1%, according to the Bureau of Labor Statistics.
What it means for career paths
The Stanford findings carry real implications for how companies structure their early-career pipelines and how recent graduates plan their professional development.
The concern is less about mass layoffs and more about a quieter reduction in the number of openings that have historically served as the on-ramp to a professional career.
That dynamic echoes concerns raised separately about how daily workplace routines are shifting for Americans across experience levels, and it adds a generational dimension that labor economists have flagged as difficult to reverse once hiring pipelines thin out for several years running.
Researchers stress that the dashboard measures the correlation between AI exposure and employment trends, not a causal link between the two. A February 2026 update noted that the decline in AI-exposed occupations only becomes statistically significant from 2024 onward.
Still, the consistent direction of the data across multiple vintages is drawing attention from economists tracking whether AI is reshaping who gets a first job before it reshapes how many jobs exist in total.
The Stanford Digital Economy Lab’s August 2026 Canaries update is publicly available alongside its broader AI Economic Indicators dashboard, which refreshes on a rolling basis using ADP payroll records. The researchers note that the gap for workers in the 26 to 30 age bracket is also showing modest decline, a sign that the pressure may be beginning to move up the experience ladder, even if the sharpest effects remain at the youngest end of the workforce.
Those career entry-point pressures align with broader questions workers are raising about what the modern job actually offers and whether showing up in person still changes the equation.
