UCLA study finds no AI-related spike in unemployment for recent graduates
New research covering summer 2026 shows the job market for young college grads looks roughly the same as it did before the AI boom.
Everyone’s been waiting for the other shoe to drop. AI was supposed to eat entry-level jobs first, turning fresh college graduates into the canaries in the economic coal mine. A new working paper from UCLA says the canaries are doing fine.
Economists Robert Fairlie and Jane Wu analyzed US labor data and found that the unemployment rate for recent graduates aged 22 to 25 during summer 2026 came in at 7.3%. That number sits comfortably within the range of the four preceding summers, which bounced between 6.3% and 7.8%. In other words, the great AI hiring freeze hasn’t materialized, at least not yet.
What the numbers actually show
The study, released as CESifo Working Paper No. 12994 and NBER Working Paper 35796, drew on US Current Population Survey microdata covering June through August 2026. The researchers chose this window deliberately: by mid-2026, workplace AI adoption was expected to be well underway, making it a natural checkpoint for labor market effects.
The year-over-year comparison is remarkably stable. Summer 2022 saw a 7.1% unemployment rate for this cohort. Summer 2023 dipped to 6.3%. Summer 2024 ticked up to 7.8%. Summer 2025 landed at 7.2%. And now summer 2026 at 7.3%.
Fairlie and Wu also checked whether graduates entering AI-exposed occupations fared worse than those in less exposed fields. The answer: no statistically significant relationship between AI exposure and unemployment rates for 2026 graduates.
AI, tech, and the markets they move—in one daily briefing.
Daily. Free. Join 34,000+ readers across crypto, finance, and policy.
They ran comparisons against two control groups, older college graduates and young people without degrees, and found no meaningful divergence there either.
An expanded unemployment measure that captures people who want jobs but aren’t actively searching pushed the figure to roughly 9.3%. That’s about two percentage points above the standard rate, which is a consistent gap across recent years rather than a new 2026 phenomenon.
The remote work wrinkle
One finding did raise an eyebrow, though it pointed away from AI. The researchers found a positive link between occupational remote-work availability and 2026 unemployment differentials. Put simply, graduates looking for jobs in fields with more remote work options appeared to face slightly stiffer competition.
The distinction matters because many AI-exposed roles are also remote-friendly roles, and sloppy analysis could attribute remote-work competition effects to AI. Fairlie and Wu’s statistical approach separated the variables, and AI exposure came up clean.
The study’s authors are careful not to declare permanent victory. Their data covers a single summer snapshot, and the long-term effects of AI on employment could look very different.