20% of workers say they use AI for tasks that used to be given to colleagues, poll finds

A new poll has found that almost half of employed U.S. adults use AI for work, with over 20% of those using the tool for work every day the poll was conducted.

The survey, by nonprofit AI research center Epoch AI in conjunction with polling firm Ipsos, also found that 1 in 5 employed adults are utilizing AI for tasks they or their team previously would have given to a contractor or co-worker.

The poll builds on previous findings from Epoch AI that artificial intelligence has become a key part of work and workflows for many employed Americans, helping workers accomplish everything from designing new software and preparing contracts to pricing goods and advising on business issues.

Amreeta Das, who helped lead the Epoch AI research, said the poll’s insight on the breadth of workers’ AI use was notable. “Our previous poll from March found that about more than half of our respondents were using AI for work,” Das said, “but we didn’t exactly know how they were using it or what kinds of tasks they were using it for.”

“People are figuring out how to divide work between themselves and the AI,” Das continued. “So far, this does not necessarily mean that there is automation of whole jobs, but there are tasks within these jobs through which AI is entering the workforce.”

Thursday’s poll found that just over 23% of respondents said in the past week they used AI to help read do***ents to inform their work, while 22% of the surveyed individuals said they used AI to help ***yze or evaluate data.

The new survey comes as American businesses wrestle with questions about how to adopt AI and stay up-to-date with the fast-moving technology. In just the past year, leading generative AI systems have transformed from primarily helping workers answer text questions and sift through data to performing substantial technical tasks, like generating the majority of computer code at major tech companies.

Other polls have found that almost half of American businesses have adopted AI, all while corporate leaders are attempting to balance AI innovation with ballooning AI budgets.

Thursday’s poll largely mirrors earlier findings from Epoch and Ipsos released in April. That poll found that roughly half of Americans used AI for work in the past week, with half of those workers using the AI systems via their workplace, compared to consumer versions of AI services.

As in Epoch’s previous polls, respondents said ChatGPT was the most popular tool used (by around 34% of employed U.S. adults), followed by Google’s Gemini (24%) and Microsoft’s Copilot (15%).

Aya Ibrahim, a senior fellow for economic and national security at the nonprofit AI Now Institute, which produces policy research on AI, and a visiting senior fellow at the progressive think tank Century Foundation, said that the poll’s task-related findings raised key questions about how AI is helping workers perform certain pieces of their job more efficiently compared to replacing entire jobs overall.

“A lot of the findings here seem to show that people are using AI to tackle some part of a larger work role or particular tasks,” said Ibrahim, who was not part of the research team but viewed preliminary results from this poll. She highlighted that the finding about delegating tasks to AI that previously would have gone to a co-worker or contractor did not necessarily mean that contractors’ roles would be replaced entirely, though task-based contractors could be more directly threatened.

Ibrahim said the poll was a gateway for holding larger conversations about the future of work and ways to create jobs that are not simply a series of replaceable tasks but instead broader portfolios of meaningful work. “These AI and tech debates are forcing mechanisms to confront issues that we have avoided for some time,” she said, highlighting the instability of entirely task-based work.

Ibrahim also stressed that the poll was a step in the right direction in that it turns to workers for direct input, instead of relying on top-down visions from corporate messengers or theoretical academic papers.

“In the discourse around AI and disruption, it feels like so much of the conversation is framed as ‘this is what’s going to happen to workers,’ and less of how workers themselves are thinking of their own precarity,” Ibrahim said.

“It’s great to see and hear more perspectives from workers on this conversation,” she said.