Stanford and ADP Data Show AI-Exposed Entry-Level Jobs Shrinking 3.8% Annually While Protected Roles Grow

[Jedi Stuart]

A dashboard tracking labour market data produced jointly by Stanford researchers and payroll processor ADP shows entry-level positions in occupations most exposed to AI automation shrinking by 3.8 percent annually among workers aged 22 to 25, compared to 2 percent growth in roles the same research classifies as least exposed to AI-driven automation. The data, drawn from ADP's extensive payroll processing dataset covering a substantial share of US private sector employment, offers one of the more granular empirical windows currently available into how AI capability is actually reshaping entry-level hiring patterns in real time, as distinct from survey-based sentiment data or anecdotal reporting.

The finding sits in tension with the supplementation-not-replacement framing that has characterised much of the AI industry's own messaging through 2026, including Anthropic's Economic Index research describing AI as predominantly enhancing rather than displacing skilled professional work. The Stanford-ADP data suggests a more complicated picture specifically at the entry-level tier, where the tasks AI most effectively automates, structured research synthesis, first-draft document production, junior-level code review and similar foundational professional skills, have historically been exactly the tasks assigned to early-career workers as both genuinely useful output and as a training mechanism for developing the judgment that more senior roles eventually require.

The age-specific framing, focused on workers 22 to 25, is methodologically significant because it isolates a cohort entering the labour market for the first time rather than mixing in the effects of AI on established mid-career or senior professionals whose roles may be more resistant to automation due to accumulated judgment, client relationships and organisational knowledge that current AI systems cannot easily replicate. If the pattern persists, it raises a structural concern beyond simple job displacement: a labour market where the traditional entry-level roles that train the next generation of senior professionals are being selectively automated away could create a skills pipeline problem that becomes apparent only years later, when there are fewer experienced professionals available because fewer people were able to enter and develop expertise through the now-diminished entry-level pathway.

Stanford Adp Canaries Dashboard Ai Exposed Entry Level Jobsaiweekly.co