Study: Highly Educated Workers Are Hit Hardest by AI
Mikhail T. (Sh0ny)
5 июля 20262 min read
In short
A report by the California Policy Lab showed that since the launch of ChatGPT 3.5 in November 2022, the brunt of AI-related layoffs has fallen on employees with higher education in the San Francisco Bay Area. Unemployment benefit claims among this group rose by about 20%.
The California Policy Lab at UCLA has published a first-of-its-kind report on the impact of artificial intelligence on California’s labor market. The study was commissioned by Governor Gavin Newsom, who in May 2026 ordered a documentation of the scale of layoffs linked to AI development.
Analysts examined data on unemployment benefit claims from the launch of ChatGPT 3.5 in November 2022 through May 2026, comparing them with the vulnerability of claimants’ occupations to AI-driven automation.
Key Findings
- The largest increase in layoffs was recorded in the San Francisco Bay Area—the epicenter of the tech industry.
- Contrary to popular belief, it was not entry-level professionals who were hardest hit, but rather experienced employees with higher education—those holding a bachelor’s degree or higher.
- In May 2026, the number of new claims for benefits from workers with high AI exposure and a bachelor’s degree or higher was approximately 4,000 higher than in November 2022—an increase of about 20%.
- Of these, about 1,700 additional claims came from bachelor’s degree holders and 2,260 from master’s degree holders.
California as the “Epicenter”
Study lead Till von Wachter called California the “ground zero” of AI development and, consequently, of AI-related layoffs. According to him, there has not been a sharp spike in layoffs at the state level—rather, there has been a gradual but steady increase.
The context confirms this trend: in May 2026, following Newsom’s executive order, Meta laid off 2,000 employees in Menlo Park. In June, Rivian cut hundreds of jobs, and Oracle announced a total reduction of 21,000 positions, attributing this to AI development. Salesforce and Elastic took similar measures.
Data Limitations
The researchers acknowledge the limitations: they are unable to accurately measure how many employees actively use AI tools in their daily work, which specific companies have implemented AI, or exactly which tools are being used. Nevertheless, the report shows for the first time, based on real-world data, that automation affects not only entry-level staff but also skilled professionals with established careers.