AI was supposed to hit new grads hard. So far, unemployment data says otherwise.

Researchers at CESifo examined U.S. labor-force survey data to assess whether accelerating workplace AI use had produced a noticeable increase in unemployment among recent college graduates in summer 2026. They report a 7.3% unemployment rate for that cohort, a level the authors describe as within the historical range for recent years and not unusually high compared with prior summers. To test robustness, the team compared recent college graduates with two alternative groups: similarly aged non-college graduates and older college graduates (ages 30–49). They also segmented outcomes by an occupation-level AI exposure metric drawn from a 2023 study that evaluated which roles were most amenable to AI technologies. Across nearly all comparisons and time-frame tests from 2022 to 2026, the researchers found no statistically significant differences that would indicate widespread displacement or reduced hiring of the 2026 graduate cohort. The CESifo paper contrasts its findings with a recent Stanford study based on payroll records from ADP. The authors note methodological differences that could explain divergent results: ADP payroll data capture the supply of jobs in various occupations, while the unemployment rate also reflects aggregate demand for work. In other words, a decline in job postings in certain fields could coexist with stable unemployment if demand shifts across the economy. The researchers characterize the Summer 2026 unemployment snapshot as an initial, informative test of AI’s labor-market effects and emphasize caution about extrapolating this single-year result into the future. They warn that if firms intensify AI adoption, later graduating cohorts—such as the class of 2027 and beyond—could experience different outcomes, and they call for additional years of data to determine whether any effects emerge as workplace AI use deepens.