Human Tech Tree
Current research2025-2026 · Present (2015 – Oct 2026)

Civilization / Society, Economy & Law

AI & the Labor Market

Early data show falling employment for young entrants in AI-exposed jobs; no economy-wide drop in employment is visible yet.

Open in the interactive tree →

The Stanford study 'Canaries in the Coal Mine?' (Brynjolfsson, Chandar, Chen; August 2025) used payroll data from ADP: workers aged 22-25 in the most AI-exposed occupations saw a 13% relative employment decline as of July 2025 data. Older workers and jobs where AI complements the work stayed stable.

As of October 2026

The August 2026 update (ADP data through June 2026) finds employment of 22-25-year-olds in AI-exposed occupations about 19% below where it would be had it kept pace with less-exposed peers, a gap that has widened since 2025 and works mainly through reduced hiring rather than layoffs. The authors find no evidence of economy-wide job displacement, see effects in employment rather than base pay, and call the results early descriptive indicators, not causal estimates. The gap persists when controlling for interest-rate exposure and remote work, but shrinks when controlling for education and partly predates generative AI, so its cause remains debated.

Open steps

  • AI-matched retraining programs Medium AI leverageBuild systems that assess displaced workers' skills, match them to retraining programs, and track outcomes.
  • Measuring automation exposure High AI leverageMap which occupations and income groups are most exposed to AI-driven job loss in real time.
  • UBI pilot outcome evaluation Low AI leverageDesign and analyse UBI pilots in high-displacement regions to measure effects on employment, entrepreneurship and outcomes.

Where AI could help

Medium AI leverage. AI speeds the measurement of which tasks are exposed, but the effects depend on firms' hiring choices and on causal identification.

  • Classify millions of job postings and usage logs by task exposure
  • Monthly dashboards from payroll microdata
  • Synthetic-control and difference-in-differences analysis at scale

Shown so far

  • In August 2026 Stanford's Digital Economy Lab reported from ADP payroll data that employment of 22-25-year-olds in AI-exposed jobs is about 19% below less-exposed peers, calling it descriptive, not causal. source
  • In January 2026 Anthropic's Economic Index report found that 52% of Claude.ai conversations augmented work and 45% automated it in November 2025, according to the company. source

Prerequisites

Unlocks

Sources

More in Society, Economy & Law · Present

All 57 points in Society, Economy & Law →

Open in the interactive tree →