Half of all US employees now use artificial intelligence at work, crossing landmark threshold for first time — Gallup data shows daily and weekly usage hitting all-time high of 28% in Q1 2026, with 65% feeling positive about its impact on productivity

Robot / AI workers at work on laptops
(Image credit: Getty Images)

Half of all American adults in active employment now use some form of artificial intelligence in their role, according to a new Gallup poll. While that figure relates to the whole number of employees using AI in some form across a year, the active number of daily and weekly users using AI has risen over the last quarter, too.

This isn’t a huge surprise: AI adoption is booming across all major industries. Major tech companies involved in AI are investing heavily in its growth, with Google owner Alphabet doubling its expenditure on AI to $180 billion this year. While a recent survey of executives by the National Bureau of Economic Research reports that over 80% of AI-consuming companies see no productivity gains from using tools to improve productivity and cut employment costs, the steamroller shows no sign of slowing down.

Latest Videos From

Gallup’s survey shows how quickly AI tools are beginning to rework the workplace around them. It also suggests that AI’s adoption is causing a large proportion of disruption to companies, with 27% of employees at companies using AI reporting a large or very large disruption to their workplace over the last year. It isn’t definitive, however, with 12% of employees at non-AI firms also reporting similar disruption.

Paradoxically, a larger number of employees at firms adopting AI report their companies both expanding with new hires (34%, in comparison to 28% against non-AI firms) and doubling down on headcount reduction (23% versus 16%). Despite the disruption, 65% of employees at companies using artificial intelligence are positive about its impact on productivity and efficiency at work, with 16% extremely positive.

It’s a double-edged sword, however. Employees who responded did suggest that AI was useful with specific tasks, like summarizing information, but the benefits don’t reach as far as improving the workplace itself. 10% of respondents said AI was having a negative impact on their work, while 21% said AI was transforming how work “gets done” at their workplace.

This survey is further proof that, so far, AI continues to grow apace, leaving organizations of all sizes struggling to adapt to its growth and adoption. What this Gallup survey suggests is that, while individual employees have begun to find specific use cases for it to help them work, there’s still work to do for AI-positive employers to successfully transform the workplace around it.

Google Preferred Source

Follow Tom's Hardware on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds.

Ben Stockton
Deals Writer

Ben Stockton is a deals writer at Tom’s Hardware. He's been writing about technology since 2018, with bylines at PCGamesN, How-To Geek, and Tom’s Guide, among others. When he’s not hunting down the best bargains, he’s busy tinkering with his homelab or watching old Star Trek episodes.

  • watzupken
    LLM is useful at work to some extent, so I think it is factual. But what is also driving up the usage number is because companies are increasingly tying use of LLM to employees' KPI. Hence, whether LLM is useful or not should be assessed by an individual's usage rather than use at work. This way, they are not compelled to use it and so a more accurate representation of actual interest.
    Reply
  • thestryker
    A lot of this could also easily be leveraging ai to deal with things like email. What exactly it's being used for is the extremely important metric this glosses over.
    Reply
  • Eximo
    Probably something like. "Have you ever used the AI summary in google"
    Reply
  • hotaru251
    specifically asking 23,717 U.S. employees

    so less than 0.006%(ish) of actual american workforce. (theres over 135M working americans)

    effectively irrelevant data size
    Reply