California has launched a new public dashboard to monitor unemployment claims related to artificial intelligence, an effort praised for its ambition but criticized for its narrow focus on job losses.
More than half of Americans express concern that AI could lead to job displacement in their households, according to a recent Reuters/Ipsos poll. In response to these concerns, the California Policy Lab, in collaboration with the state's Employment Development Department, created the dashboard to track AI-related unemployment claims in near real-time. This tool joins other resources like the Stanford Digital Economy Lab dashboard for analyzing AI's impact on the workforce.
However, the California tracker faces significant limitations. Primarily, it measures only one side of AI's labor market impact: job losses. It does not capture jobs created by AI, the workers it helps retain, or productivity gains that may lead to increased hiring. This sole focus on unemployment risks presenting an incomplete picture of the labor market's evolution.
Further complicating the data is California's reliance on the Dictionary of Occupational Titles, a system last updated comprehensively in 1991. Modern roles like data scientists and machine-learning engineers are absent from this outdated classification. Researchers must translate these obsolete codes into modern measures of AI exposure, which can average together workers whose current jobs may differ significantly from historical descriptions.
The dashboard's primary finding—an increase in unemployment claims among workers in AI-exposed occupations starting in late 2022—coincides with the public release of ChatGPT. Yet, this period also marked the end of the pandemic-era technology hiring boom, with companies like Meta, Google, and Amazon announcing significant layoffs amid rising interest rates. Many occupations highly exposed to AI were also heavily impacted by this market correction, making it difficult to isolate AI's specific impact from broader economic trends using unemployment claims alone.
Even if all recorded joblessness were attributed to AI, unemployment claims only account for separations and do not reveal what happens to displaced employees afterward. Research using administrative data across the U.S. indicates that industries with higher AI exposure have experienced faster productivity growth, increased employment, and higher wages through early 2024. Other studies also show surprisingly modest evidence of widespread labor disruption from large language models, with some analyses of company spending records showing that firms investing heavily in AI expanded employment more rapidly.
Previous technological revolutions have followed a pattern where some jobs disappear, many change, and new opportunities emerge alongside increased productivity. A dashboard that only counts layoffs fails to illustrate this dynamic process. For instance, an "Automobile Job Loss Tracker" initiated in 1910 would have accurately documented the decline of buggy-whip makers but would have missed the burgeoning roles of mechanics, drivers, and engineers created by the automobile industry.
California possesses existing data, such as quarterly wage records covering nearly every employer, which could reveal if displaced workers find new jobs, how their earnings change, and which industries are expanding. Combined with modern job-posting data, these resources could identify growing demand for AI-related skills across different occupations, industries, and regions.
To further improve insight, California could modernize its reporting requirements. Adopting the federal Standard Occupational Classification system or requiring employers to report actual job titles on quarterly wage records would provide policymakers with a clearer understanding of how work is evolving. Providing workers with information on emerging opportunities, in-demand skills, salary expectations, and qualification pathways would be more beneficial than simply tracking layoffs. California has an opportunity to not only measure disruption but also to build an "AI opportunity tracker" to help workers navigate the changing labor landscape.