California has started to track AI-related job loss

The California Employment Development Department launched an online data tool late last month that aims to help state leaders and other stakeholders address how the workforce could be impacted by the growing adoption of artificial intelligence. 

The California AI-Unemployment Tracker — or CAIT — aims to offer oversight on labor market changes within the state that could be associated with AI’s use in workplaces. The tool considers unemployment insurance claims data and exposure metrics, such as whether a certain job includes tasks that can be done with AI. 

“There is considerable speculation about how AI may affect job loss, but very little evidence. This new tracker is a way to add some evidence to the conversation …. and identify potential signs of disruption as they emerge,” said Ben Hyman, a senior researcher at the California Policy Lab, which partnered with the EDD to build the tool and publish an accompanying report with preliminary data insights last month. 

But it’s imperative for policymakers and other users to consider that “these findings are best understood as an early signal, not a call for any particular policy response,” Hyman, a co-author of the report, said. “The tracker helps policymakers understand what’s changing and make informed decisions as more evidence becomes available.” 

In fact, there is “no evidence of rising UI claims from AI-exposed occupations” between January 2019 and May 2026, according to the report. 

While there is no significant evidence of AI-related mass layoffs across the state thus far, “we did see some patterns in certain regions like the Bay Area, in certain tech-heavy sectors and among highly AI-exposed workers with college degrees, so we’ll continue to monitor those trends,” Hyman said. 

For instance, UI claims among college-educated people in jobs with high-risk exposure to AI increased from an average of 13,000 claims per month in November 2022 — the month ChatGPT was released to the public — to between 16,000 and 22,000 claims per month since 2023, according to the report. 

Still in its early stage of use, CAIT will continue to be updated every month with UI claims data and developers will consider adding more data points to the dashboard in the future, Hyman said. 

“We hope the dashboard and report can be illuminating for policymakers in other states who are also thinking about the potential impacts that AI may have on workers and how they may respond,” he said. “This is a topic that’s at the forefront of many conversations, especially as AI continues to become more integrated into more sectors.”  

Indeed, humans tend to focus on “what we can measure and what we can understand,” said Jeffrey Wenger, senior economist at RAND. For instance, people can observe that AI is increasingly assuming traditionally manual work, like data entry or drafting communications, “and then we immediately interpret that as resulting in reduced employment,” he explained. 

He pointed to the fact that, around the same time the report identified spikes in UI claims, the U.S. was still recovering from COVID-19 pandemic-related disruptions that also likely contributed to rising unemployment cases. Factors like a nationwide recession and spikes in work-from-home orders also impacted people’s job statuses. 

While data initiatives like CAIT can be valuable, particularly as more information is revealed with time, policymakers should resist the urge to draw strong links between AI and certain outcomes, he said. 

Indeed, the report states “... our measures of AI exposure capture the extent to which job tasks could be or have been performed by AI, not necessarily whether AI is being adopted or causing displacement at a specific workplace. It is also well known that UI claims capture only workers who claim UI benefits after losing their jobs, rather than all people who lose their jobs.” 

policymakers should remember that “there are potentially many, many other jobs that [AI] will create and generate new opportunities [for],” Wenger said. 

A growing number of state and local governments are creating roles like a chief AI officer, to oversee the exploration, research and implementation of the technology. Illinois appointed its first CAIO last month, joining a handful of states, including Alabama, North Carolina, Oklahoma, Montana and Texas, that have already adopted the position, GovTech recently reported. 

Existing government leadership roles, like the chief information officer, have also been expanding from primarily managing technology infrastructure to now include responsibilities like driving innovation and modernization government- or agency-wide as AI is used more. 

And many states and localities have increasingly launched workforce upskilling and development programs intended specifically for AI-oriented jobs and job functions. 

Ultimately, “we can look at the amount of disruption [of AI’s impact on the workforce] and be a little bit concerned, but I think we have to look at … the ability of the people who are disrupted to adapt to it,” Wenger said.