AI's Impact on California Jobs: Nuanced Findings from the New AI-Unemployment Tracker
In a sea of alarming headlines about AI-driven layoffs at major tech companies, the California Policy Lab is offering a more balanced perspective. According to the research institute, the reality is nuanced—though cold comfort for Bay Area-based, college-educated tech workers.
A new **California AI-Unemployment Tracker** from the lab reveals that AI has not triggered the widespread job losses once forecasted by executives at companies like Anthropic, Amazon, and Ford. However, early warning signs are emerging for roles deemed “highly exposed” to AI.
The data shows that college-educated workers (including those with master’s and Ph.D. degrees) in high-exposure occupations—such as software developers and customer service representatives—experienced a spike in unemployment claims of more than 50% within a year of ChatGPT’s public release in late 2022. Claims have since remained elevated, though at lower levels. By contrast, similarly educated workers in low-exposure fields like nursing and agriculture saw far fewer claims.
The impacts are most pronounced in the Bay Area, where highly AI-exposed jobs are experiencing some of the largest employment disruptions. Notably, unemployment claims in this group were already consistently higher even before ChatGPT’s debut.
The tracker, designed to help policymakers anticipate and respond to labor market shifts, merges monthly unemployment insurance claims from the California Employment Development Department with AI exposure metrics developed by OpenAI and Anthropic. “California is the first state to do this,” said Ben Hyman, a UCLA economist who co-authored the study and helped build the tool. “I expect other states to follow.”
**The Standard** spoke with Hyman about the findings and what the tracker reveals about the jobs AI is beginning to disrupt.
**The Standard:** It’s striking that the Bay Area—the epicenter of AI innovation—also appears to be where the technology is hitting hardest.
**Ben Hyman:** Occupations that are highly exposed to AI, particularly in the Bay Area’s tech and tech-related industries and among workers with advanced degrees, are seeing more persistent new unemployment insurance claims. These are exactly the areas where we’d expect to see effects first.
We’re observing some job losses among certain software engineers, but we’re not seeing a corresponding surge in hiring for engineers who are more AI-native or deeply integrated with these new tools.
From a policymaker’s perspective, the focus is understandably on the negative impacts. The tracker’s strength is its ability to detect large spikes or trend changes across other industries, which we’ll continue monitoring closely.
**The Standard:** Since tech jobs and the Bay Area are being affected first, are they likely to be hit the hardest overall?
**Ben Hyman:** That’s the million-dollar question—what can we learn from tech that might apply to other sectors and the future of the industry itself? Our data shows an early surge in the information sector, which has since returned closer to baseline unemployment claims. Meanwhile, professional services—a sector mixing tech and customer service roles—has remained relatively stable.
It’s difficult to project forward. Our lab doesn’t focus on long-term forecasting. Still, our findings align with national analyses from the Yale Budget Lab and a recent Anthropic report, both of which have found no major effects on unemployment so far.
**The Standard:** How should we view the more dramatic predictions about AI replacing jobs? In San Francisco, there’s a palpable sense of fear in tech circles.
**Ben Hyman:** The tracker helps explain why perceptions differ so sharply. If you’re focused on the Bay Area, tech industries, and workers with advanced degrees, the changes feel very disruptive—because that’s where the clearest signs of labor market pressure appear. But when you zoom out to the full California economy, those patterns haven’t spread widely.
**The Standard:** Where are policymakers focusing, and what are their main takeaways?
**Ben Hyman:** They’re thinking about how to support workers displaced by AI. Should they prioritize reskilling? Lean into AI tools? Or shift toward historically resilient fields like nursing and medical assisting, which have lower exposure?
If these effects become more widespread, we may need more creative, targeted approaches to unemployment insurance. Beyond policymakers, we hope individuals will use the tool too—whether it’s historically disadvantaged groups examining impacts on their communities or undergraduates deciding whether to pursue a computer science degree or pivot to another field.
**The Standard:** What remains to be done in this research area?
**Ben Hyman:** The critical missing piece is direct data on what happens *after* employers adopt AI tools intensively. Linking that adoption data to labor market outcomes will be transformative once integrated into these rich datasets.
The California AI-Unemployment Tracker provides a valuable, data-driven lens into an evolving labor market. While a broad catastrophe has not materialized, the tool highlights real pressures in specific pockets—particularly for highly skilled tech workers in innovation hubs. As AI adoption accelerates, this kind of granular tracking will be essential for guiding policy and personal career decisions alike.