What is really happening to jobs? Separating AI hype from reality | Stanford Institute for Economic Policy Research (SIEPR)
siepr.stanford.edu · 5 November 2026
Among firms that adopted enterprise AI, employment grew by 10 percent in the two years following adoption, an effect driven by firms with the highest per capita AI spending. Unemployment among workers in occupations most exposed to AI-driven disruption has risen by 0.77 percentage points since 2022, compared to 0.85 percentage points for least-exposed workers. Employment trends in occupations with high exposure to AI are fairly stable. Job postings for software developers have been growing faster than for other occupations over the last year.
This is the observable fact the cited source supports, stated without interpretation.
Tests COMMERCIAL_ALIGNMENT and REVENUE_DURABILITY by examining whether autonomous deployment correlates with changes in employment structures and job posting activity. If firms adopting AI at scale show employment contraction despite capability gains, that would contradict the hypothesis that value capture can reorganise through structural shifts. The fact that employment grew, especially in high-spending firms, is consistent with a transition phase in which activity and value are both being reshaped.
The framework anticipates that autonomous capability deployment can coexist with employment growth where value capture restructures rather than contracts immediately. The observed 10 percent employment growth in firms with highest per capita AI spending is consistent with commercial alignment occurring during a transformation phase. The fact that most-exposed occupations have not experienced faster unemployment than least-exposed occupations contradicts simple displacement narratives. However, the observation that HR executives report role consolidation and hiring avoidance in roles where AI automates tasks suggests value is being recaptured through reallocation of labour rather than elimination of it.
The framework anticipates that autonomous capability deployment can coexist with employment growth where value capture restructures rather than contracts immediately. The observed 10 percent employment growth in firms with highest per capita AI spending is consistent with commercial alignment occurring during a transformation phase. The fact that most-exposed occupations have not experienced faster unemployment than least-exposed occupations contradicts simple displacement narratives. However, the observation that HR executives report role consolidation and hiring avoidance in roles where AI automates tasks suggests value is being recaptured through reallocation of labour rather than elimination of it.
This is the framework's reading of the fact above. The source does not endorse it and is not cited as doing so.
The text reports empirical aggregates as of 2026. The 10 percent employment growth in adopting firms covers two years following adoption but does not establish when adoption occurred relative to the April 2026 paper publication. This is early in the relevant test window for commercial durability (1 to 2 years post-deployment). It shows employment structure changing rather than contracting, consistent with value reaccumulation, but does not yet establish whether that reaccumulation is moving toward orchestration, verification and governance (LAW_1) or remaining in activity-unit economics.
Current evidence is recorded immediately. The paper's longer empirical horizon determines when persistence becomes a stronger framework-level test.
Employment growth in adopting firms could reflect timing selection bias if firms adopted AI during a growth phase, or investment in parallel capability rather than autonomous substitution. The comparison of unemployment rates across exposure quintiles does not account for compositional shifts in who works in exposed occupations or for wage-level impacts on job-seeker behaviour. The text emphasises that HR executives report hiring avoidance in automatable roles, which suggests a different mechanism than measured job losses: employers are not filling vacancies that would have been filled before, a form of value capture that does not appear in employment headcount.
Recorded for every observation assessed at strength 4 or 5. A record that cannot state the strongest competing reading of its own evidence is not published.
Where several outlets report the same development, they are recorded as additional sources on this one observation and are not counted as separate evidence.