American workers adopted artificial intelligence into their jobs at a rapid pace over the past few years, according to a Gallup Workforce survey of more than 22,000 U.S. workers conducted in late October and early November 2025. Twelve percent of employed adults said they use AI daily on the job, while roughly one-quarter reported using it at least a few times a week, and nearly half said they use it at least occasionally.
The sharp increase from just 21% occasional AI users in 2023 reflects the commercial boom that ChatGPT, Google’s Gemini, and other generative AI tools sparked after their introduction. Yet adoption remains uneven: while technology, finance, and education sectors lead with majorities of workers using AI regularly, research identifies about 6.1 million U.S. workers who are heavily exposed to AI disruption but lack the skills and financial cushion to adapt.
AI adoption in the American workplace is accelerating, with workers across sectors increasingly turning to chatbots and generative AI tools to perform their jobs more efficiently.
Twelve percent of employed adults said they use artificial intelligence daily on the job, according to a Gallup Workforce survey conducted October 30 through November 13, 2025. Roughly one-quarter use AI at least a few times a week, and nearly half reported using it at least occasionally.
The figures mark a sharp increase from 2023, when just 21% of workers reported any regular AI use. The surge reflects the commercial explosion that ChatGPT, Google’s Gemini, and other generative AI tools sparked after their introduction, placing AI squarely in workplace workflows across industries.
High adoption in tech, finance, and education
Technology workers are leading the adoption curve. About 60% of technology sector employees use AI frequently, and roughly 30% do so daily. The share of technology workers using AI daily or regularly has grown substantially since 2023, though some signs suggest adoption may be starting to plateau after explosive growth between 2024 and 2025.
Finance and education sectors also show high adoption rates. Andrea Tanzi, a 28-year-old investment banker at Bank of America in New York, said he uses AI tools daily to synthesize documents and data sets that would otherwise take several hours to review. He also relies on the bank’s internal AI chatbot, Erica, to handle administrative tasks.
In education, majorities of workers in K-12 schools, colleges and universities, and professional services report using AI at least a few times a year. Joyce Hatzidakis, a 60-year-old high school art teacher in Riverside, California, started experimenting with AI chatbots to draft communications with parents. “I can scribble out a note and not worry about what I say and then tell it what tone I want,” she said. “And then, when I reread it, if it’s not quite right, I can have it edited again. I’m definitely getting less parent complaints.”
Hatzidakis began with ChatGPT but switched to Google’s Gemini when her school district made that its official tool. She has even used it to help draft recommendation letters, noting that “there’s only so many ways to say a kid is really creative.”
Chatbots drive the adoption
Among AI users at work, about 60% rely on chatbots or virtual assistants when they turn to AI tools. Roughly 40% reported using AI to consolidate information, generate ideas, or learn new things.
Service sectors—retail, healthcare, and manufacturing—show lower adoption rates. Gene Walinski, a 70-year-old Home Depot associate in New Smyrna Beach, Florida, adopted AI on his own initiative. He uses an AI assistant on his personal phone roughly every hour during his shift to answer customer questions about supplies. “I think my job would suffer if I couldn’t because there would be a lot of shrugged shoulders and ‘I don’t know’ and customers don’t want to hear that,” he said.
Home Depot did not ask him to use AI when he started last year, after a decades-long career in the automobile business, but the company has not discouraged it either. Walinski said he is unconcerned about AI replacing him. “The human interface part is really what a store like mine works on,” he said. “It’s all about the people.”
A different picture for vulnerable workers
Yet the rapid spread of AI at work is creating uneven impacts across the labor market. Research by Sam Manning, a fellow at the Centre for the Governance of AI and co-author of analyses for the Brookings Institution and the National Bureau of Economic Research, identifies about 6.1 million U.S. workers who are heavily exposed to AI but poorly equipped to adapt if their jobs are disrupted.
Most of these vulnerable workers are concentrated in administrative and clerical positions, are approximately 86% women, and are older and concentrated in smaller cities with fewer job alternatives, such as university towns or state capitals. “If their skills are automated, they have less transferable skills to other jobs and they have lower savings, if any savings,” Manning said. “An income shock could be much more harmful or difficult to manage.”
Workers most highly exposed to AI—those in computer-based jobs involving heavy AI usage—generally have characteristics that make them more adaptable. “Most of the workers that are most highly exposed to AI, who are most likely to have it disrupt their workflows, for good or for bad, have these characteristics that make them pretty adaptable,” Manning said. “They usually have higher levels of education, wider ranges of skill sets that can be applied to different jobs, and they also have higher savings, which is helpful for weathering an income shock if you lose your job.”
Few workers fear replacement
Despite the growing presence of AI in workplaces, few workers express concern about job replacement. A separate 2025 Gallup survey found that half of employees said it was “not at all likely” that new technology, automation, robots or AI would eliminate their job within five years. That represents a decline from about 60% who said the same in 2023.
Rev. Michael Bingham, pastor of Faith Community Methodist Church in Jacksonville, Florida, is similarly skeptical about AI’s utility in his work. When he asked a chatbot about the medieval theologian Anselm of Canterbury, it gave him “gibberish.” He said he would never ask a “soulless” machine to help write his sermons. “You don’t want a machine, you want a human being, to hold your hand if you’re dying,” he said. “And you want to know that your loved one was able to hold the hand of a loving human being who cared for them.”
The Gallup Workforce survey queried a random sample of adults age 18 and older who work full time and part time for organizations in the United States. The most recent survey of 22,368 employed U.S. adults was conducted October 30–November 13, 2025. The margin of sampling error for all respondents is plus or minus 1 percentage point.