Wire flash
St. Louis Fed Study: AI adoption widespread but not yet replacing jobs
Editorial responsibility
- No named human review is recorded for this page.
- Source reporting is collected, normalized, translated or condensed automatically when needed.
- Automatically published source-backed update
A new study by the Federal Reserve Bank of St. Louis, funded by the Washington Center for Equitable Growth and the Russell Sage Foundation, surveys AI adoption in the workplace. The research finds that AI use, particularly ChatGPT, is surprisingly widespread across 75% of occupations, with at least a fifth of workers in those fields using it. However, adoption is not deep; most workers use AI for only a few tasks, and no task has been overwhelmingly taken over by AI. The most common uses include reading documents, preparing reports, and analyzing data, with adoption rates around 60%. Productivity gains are self-reported at about 5%. Contrary to expectations, healthcare professionals have not widely adopted AI for diagnosis, while some manual labor jobs have seen higher uptake for research purposes. Researcher Adam Blandin notes that understanding personal differences in AI uptake is underexplored. The article suggests that, like computers historically, AI may complement workers rather than replace them, though the long-term impact remains uncertain.
Source report
Artificial Intelligence is poised to destroy the world—or humanity at least. In the meantime, it's solving unsolvable math problems, possibly with dirty tricks; spying on activists; and stealing our jobs. At least, that's what recent headlines suggest.
But actual research on the real-world adoption of the myriad tools lumped under the AI label—especially large language models like Claude or ChatGPT—has been slim. That is slowly beginning to change.
Key Findings from the St. Louis Fed Survey
The Federal Reserve Bank of St. Louis, with a joint grant from the Washington Center for Equitable Growth and the Russell Sage Foundation, has been investigating who is using AI, in what fields, and for what tasks. The results from their latest survey, released this month, complicate the picture of how AI is being integrated into the workplace—rapidly, yes, but not yet deeply.
Widespread but Shallow Adoption
- In 75 percent of occupations, at least a fifth of the workforce used AI.
- ChatGPT is the most commonly used AI program.
- "That's kind of just a wider distribution than we were expecting," said Adam Blandin, a St. Louis Fed research fellow and assistant professor of economics at Vanderbilt University. "Everybody expects computer programmers or executives or something like that to use it, but it's just more widely spread across occupations than we realized."
Despite broad adoption, most people use AI for only a small portion of their jobs. No single task defined by the survey had been overwhelmingly taken over by AI. Even the tasks for which AI is most commonly used—reading documents for technical information, preparing research reports, and analyzing data to identify trends—had adoption rates of about 60 percent.
Patterns and Surprises
Some patterns made intuitive sense:
- Computer programmers use AI frequently.
- People who use AI often at work are more likely to use it in everyday life.
Other findings defied expectations:
- Health care professionals have not widely adopted AI for diagnosis or treatment planning, possibly due to concerns about trust and sensitive information.
- Hands-on jobs—such as testing materials for environmental hazards or repairing machines—have seen higher-than-expected uptake. Workers in these roles use AI for research on tools and repair methods.
"Understanding those personal differences, I think, is underexplored," Blandin said.
Productivity Impact
Workers who use AI reported that their tasks would have taken 5 percent longer without it—a self-reported estimate that comes with a significant caveat. Still, the finding suggests workers are using AI because it makes them marginally more productive, not because it can replace them entirely.
What This Means for Workers
The researchers draw a parallel to the adoption of computers: when workers adopted computers, jobs relying on them became more in demand, and computer users began earning more than non-users. If AI proves similarly useful for workers—but not as workers—it could follow the same pattern.
However, one major difference remains, and the full implications are still unclear.
Source
The New RepublicWestern