Trends›AI augments more than it automates
AI augments more than it automates
71 ↓2 confidence, computed 23 Oct 2026
- Social
What rests on thismarkets it underwrites or threatens
Threatens
Evidencethe schedule the figure draws
The evidence, newest first
- Oct 2026
against · observable UK 16-24 unemployment rose 14.2% to a 16.4% peak (15.6% Jun-Aug 2026) but the rise sits in retail/hospitality (-117k payroll) and 16-17s; 18-24 employment began falling in 2022, before ChatGPT; legal/accounting hiring intentions above average (CIPD +17). Resolution Foundation + ONS LMS Oct 2026 (UK youth unemployment) - Oct 2026
for · observable US 16-24 unemployment 9.9% Sep 2026 vs 9.6% Apr 2025 (revised); prime-age flat 3.6%, U-6 7.6%. BLS Employment Situation Sep 2026 (youth 16-24 SA) - Oct 2026
against · observable Cross-country youth unemployment tracks local cycles not AI adoption (Spain -2.7pt, Finland +3.8pt, Netherlands flat); ECB finds euro-area youth employment broadly cyclical, with young ICT employment -18.6% the AI-shaped outlier it declines to attribute. Eurostat youth unemployment Aug 2026 + ECB Economic Bulletin 5/2026 box - Sep 2026
for · observable Texas admin data, flat pre-trends - 10pt higher automatable-task share in a major lowers first-year employment 1.7pt and earnings ~5%. Dodini & Smith, Dallas Fed — AI plays a role in weak labor market for college graduates - Sep 2026
neutral · observable Junior share at AI-adopting multinational affiliates falls 1.9pt, but mostly because seniors rise 6.7%; junior headcount -2.5% n.s.; total employment +3.3%. Chandar & Klein Teeselink, How Does AI Change Labor Demand? Evidence from 41 Countries (SSRN 7498743) - Aug 2026
for · primary 22-25s in the most AI-exposed jobs are ~19% below counterfactual (up from ~15% a year earlier), via reduced hiring not separations; authors now call it descriptive, concede pre-trends and attenuation with education controls. Canaries in the Coal Mine, Aug 2026 revision (ADP payroll to Jun 2026) - Aug 2026
for · observable Recent-grad unemployment 5.70% vs all workers 4.06% (gap +1.64pt, 4th highest on record; record +1.78 was Apr 2025); underemployment 41.95% after 42.37% Nov 2025 peak; young non-grads 7.19%. NY Fed Labor Market for Recent College Graduates (Q2 2026) - Jun 2026
for · primary For workers ages 22 to 25, employment in highly AI-exposed occupations is now shrinking at 3.8% per year as of April 2026. 'It's not going away': The Stanford economist who called the AI entry-level jobs crisis early has the receipts - Jun 2026
for · observable Youth unemployment stood at 9.4% in May 2026, while the employment-population ratio dropped to 53.1% from 54.5% in July 2025, indicating fewer young people working. Youth (16-24) Unemployment Rose to 9.4% in May 2026; Employment-Population Ratio Fell - Jun 2026
for · observable Employment for software developers aged 22-25 fell nearly 20% since 2024 — the cohort entering as generative-AI coding tools became standard. Stanford AI Index 2026 — 22-25yo developer employment ~-20% (coverage) - Jun 2026
for · primary Private source PwC 2026 Global AI Jobs Barometer (1 billion+ job ads analyzed) - Jun 2026
for · primary As of May 2026, the U-6 unemployment rate stands at 8.1%, sitting more than 1 percentage point above pre-pandemic norms and significantly higher than U-3 headline unemployment at 4.3%. U-6 Unemployment Rate at 8.1% in May 2026, Above Pre-Pandemic Trends - Jun 2026
for · observable Entry-level IT job share fell from 8.1% to 7.4% YoY while senior postings rose from 38.8% to 43.1%, and general software-engineering postings remain ~49% below the Feb 2020 baseline — ladder-bottom compression. Entry-level tech share falling, senior rising (Indeed via TechTimes) - Jun 2026
against · observable Remote work explains ~64% of the rise in unemployment for graduates under 29 (2017-19 vs 2022-25); AI exposure does not explain the divergence (remote-feasibility and AI exposure correlate ~0.7). NY Fed - Remote Work Leaves Younger Workers Sidelined (press coverage) - May 2026
against · observable Finds the 2026 new-grad market weaker but 'more mixed than headlines suggest,' cautioning cyclical softness and post-pandemic normalization, not AI, explain much of the entry-level weakness. EPI — Class of 2026 weaker but mixed picture - May 2026
for · primary Within-firm design (firm x quarter FE, no pre-trend) finds junior employment at GenAI adopters ~9% lower within six quarters, seniors unchanged, via slower hiring. Hosseini & Lichtinger, Generative AI as Seniority-Biased Technological Change (SSRN 5425555, May 2026 version) - May 2026
for · observable Recent-graduate unemployment stayed elevated at ~5.7% in Q1 2026 with underemployment at 41.5% (42.5% in Q4 2025, highest since 2020) — elevated underemployment without a headline jobless spike. NY Fed recent-grad unemployment ~5.7%, underemployment 41.5% - May 2026
for · primary Employment among software developers aged 22–25 has plummeted nearly 20% since 2024, even as their older colleagues' headcount grows. Inside the AI Index: 12 Takeaways from the 2026 Report - Apr 2026
for · primary Entry-level hiring collapses while AI hiring explodes; routine operational tasks at junior levels are increasingly automated or outsourced. Tech hiring trends in 2026: the 4 big shifts shaping the tech job market - Apr 2026
for · primary 42.5% of recent college graduates were underemployed in Q4 2025, the highest rate since 2020, per Federal Reserve Bank of New York. Recent College Graduates Hit 42.5% Underemployment in Q4 2025, Highest Since 2020 - Apr 2026
against · primary Goldman Sachs analysis shows AI substitution is erasing 25,000 jobs/month but augmentation adds back 9,000, with widened unemployment gap for entry-level in AI-exposed roles rather than headline spike Goldman Sachs U.S. Daily: AI employment research (substitution vs. augmentation dynamics) - Mar 2026
neutral · observable Openings for routine roles fell 13% post-ChatGPT while analytical/technical jobs grew 20%, showing reshaping rather than elimination and supporting AI-augmentation-first thesis over massive displacement Harvard Business Review: How AI Is Changing the Labor Market (job postings analysis 2019-2025) - Feb 2026
against · observable Job-finding rates among recent college graduates have been declining and unemployment gap with high school graduates narrowing, not suggesting AI-driven job destruction. Are Young College Graduates Losing Their Edge in the Job Market? - Feb 2026
against · primary Found no detectable economy-wide disruption in the ~33 months since ChatGPT: occupational/industry dissimilarity and unemployment-by-AI-exposure all flat or within historical ranges. Yale Budget Lab — no economy-wide AI disruption (Nov/Dec CPS) - Jan 2026
against · observable Concludes aggregate labor data show no AI-driven employment shock yet — supporting the 'deferred to 2027-2030' timing but undercutting any claim the effect is already large. Brookings — no AI jobs apocalypse, for now - Nov 2025
for · observable Young workers are struggling not because entry-level jobs disappeared but because employers are not expanding workforce in low-hire environment. Indeed 2026 US Jobs & Hiring Trends Report - Nov 2025
for · primary Employment declines are concentrated in occupations where AI is more likely to automate, rather than augment, human labor. Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence - Stanford Digital Economy Lab - Aug 2025
for · primary Found a 13% relative decline in employment for early-career workers (22-25) in the most AI-exposed jobs since late 2022, concentrated in automatable occupations; augmentation-heavy roles showed no comparable decline. Brynjolfsson et al. (Stanford/ADP) — Canaries in the Coal Mine
What would move itthe measures watched
- watchNY Fed recent-college-graduate unemployment & underemployment. Rising grad unemployment AND underemployment is the signature of 'hidden underemployment' rather than a headline spike
- watchEntry-level vs senior share of job postings (esp. software/IT).
- watchStanford/ADP payroll microdata — young-worker employment in AI-exposed jobs. A relative decline for 22-25yos concentrated in automatable occupations supports automation-first, junior-first
- watchBLS youth (16-24) unemployment & U-6 underemployment. A headline youth-unemployment spike would CONTRADICT the 'hidden not headline' framing; belief predicts elevated U-6 without a dramatic U-3 break
- watchCensus BTOS production-use AI rate by function. Belief requires augmentation-first — watch whether AI use stays assistive vs spreading to wholesale headcount replacement