By Megan Leonhardt
One of the big goals of artificial intelligence is to improve efficiency. But workers who regularly use generative AI actually feel less productive, new data show.
Workers who use AI tools like ChatGPT, Midjourney, Copilot, or similar programs on a daily basis were four times more likely than nonusers to say they were feeling less productive than they could otherwise be, according to a new survey conducted by payment processor ADP.
That may sound counterintuitive, but ADP's chief economist Nela Richardson says the findings make sense, in part because most workers are using AI to complete the day-to-day routine tasks -- such as scraping emails, responding to messages and summarizing documents -- that typically make workers feel productive. The elimination of such tasks can leave a gap that may call for a shift in how productivity is considered and measured, Richardson added.
Moreover, workers may also have outsize expectations around AI's automation capabilities, with many of the current tools still requiring workers to double check the completed work for errors.
The latest ADP survey data -- spanning 39,000 workers in 36 countries -- supports other data that show generative AI spend is rising, but direct productivity gains from the technology aren't quite materializing yet. Nonfarm business-sector labor productivity rose by just 1.8% in the fourth quarter, according to the latest estimate of the Bureau of Labor Statistics.
Annual average productivity increased 2.1% from 2024 to 2025. That's significantly above the trend of 1.5% growth experienced during 2007 through 2019, but still slightly lower than the long-term rate of 2.2% the U.S. has seen since 1947, the BLS reported.
Many economists believe that recent gains in productivity have been driven by investment, changing labor-market dynamics, and business dynamism, rather than the direct effects of AI. Yet productivity gains, while volatile, are important to watch carefully as generative AI gains scale because they can drive economic growth and help to alleviate inflationary pressures.
That said, history has shown that technological innovations like AI can reduce productivity growth at first, as companies invest time and resources in learning and developing processes around the latest improvements, before the benefits accrue years or even decades later.
It also may still be too early to see huge efficiency gains, however, because AI hasn't reached widespread adoption. Overall, about 20% of workers globally say they use AI nearly every day. About 30% of workers report using it multiple times a week, while about 15% of workers still haven't tried out the emerging technology.
Among U.S. workers, about 19% report using AI on a daily basis, but more, 23%, report never having picked it up.
That tracks closely with census data that found about 18.2% of businesses reported using any type of artificial intelligence -- including machine learning, natural language processing, virtual agents and voice recognition -- in their business operations in February.
In the ADP survey, about 16% of daily users said they felt less productive at work than they otherwise could be, as opposed to just 5% of workers who felt less efficient and report never using AI.
Despite the lack of noticeable productivity gains, ADP's survey did find that workers who regularly used AI were more engaged, less stressed, and more positive about their colleagues. These workers also reported feeling notably more confidence in their job security.
That could be a tangible benefit considering ADP found that only 22% of global workers believed their job was secure. The U.S. workforce was slightly more confident: about 28% of Americans felt confident their job would not be eliminated.
"Job security doesn't just matter for the job, though. People who are more secure in their jobs also feel more financially secure," Richardson said. "There are ripple effects between what we do at work and what we do at home, and those of sentiment and security."
Write to Megan Leonhardt at megan.leonhardt@barrons.com
This content was created by Barron's, which is operated by Dow Jones & Co. Barron's is published independently from Dow Jones Newswires and The Wall Street Journal.
(END) Dow Jones Newswires
March 25, 2026 11:25 ET (15:25 GMT)
Copyright (c) 2026 Dow Jones & Company, Inc.
Comments