How are Americans using AI? Evidence from a nationwide survey

Introduction

The rapid emergence of Artificial Intelligence (AI) technology has heightened the need to better understand its adoption across various aspects of social and economic applications. In this essay, we present new evidence on the extent of AI adoption across a host of dimensions, including use in households, by employees in the workplace, and by owners and workers in small businesses.

To achieve a credible, nationally representative sample, we added questions on AI utilization to the AmeriSpeak Omnibus survey conducted by the National Opinion Research Center (NORC) at the University of Chicago. Because the survey panel contains rich demographic information, we are able to better understand AI usage across age, income, and racial categories, in addition to occupation and size of employer.

The survey results suggest widespread use of generative AI, with 57% of respondents using AI for personal purposes and 40% reporting an increase in their use of the technology over the past year. Across the survey, higher levels of education were strongly associated with increased AI usage. About one-in-five respondents use AI in their professional lives, although the share is sharply increasing by education and income. Workers are also somewhat pessimistic about AI and labor market displacement, with more believing that AI will reduce the number of jobs than believing it will increase opportunities. Lastly, usage patterns by firm size are strikingly similar, as use in small businesses is virtually identical to that in large ones. In sum, while AI use is widespread across personal and professional lives, there remain abundant opportunities to expand take-up, in particular among workers and households with lower levels of education.

The survey results lend themselves to three main conclusions: One, AI use is remarkably consistent across firm size. Two, while personal use is common, professional AI use is far from ubiquitous and many respondents expressed skepticism that it would be as revolutionary as some experts expect. And three, there are important differences across AI use by demographics, including increased use among those with higher education and lower usage for retirement-age respondents.

Prior surveys on AI usage

In recent years there have been numerous studies examining AI usage trends at the personal and household levels, as well as in a variety of American industries. The sectors that have been examined most frequently include health care, finance, and government. Collectively, these studies suggest that the rate and nature of AI adoption varies a great deal across sectors, as well as between firms within sectors.

On the individual level, Bick, Blandin, and Deming (2025) found that, as of late 2024, 39.6% of Real-Time Population Survey respondents aged 18 to 64 used generative AI. A 2025 YouGov survey found that 56% of American adults used AI tools, and 28% use AI tools at least once per week. These rates were even higher among adults under 30, for whom the rate of general use was 76% and weekly use was 50%. A survey conducted by the Pew Research Center found that 57% of surveyed U.S. adults said that they interacted with AI at least several times a week, though only 33% said they have ever used an AI chatbot. The individual use of AI is potentially even higher than these surveys estimate, as a study conducted by Gallup and Telescope in late 2024 revealed that, while approximately 99% of Americans use at least one product with AI features weekly, only 36% of respondents realize that these products use AI. These AI-enabled products include personal virtual assistants and navigation apps.

In the business sector, McElheran et al. (2024) concluded that less than 6% of the 850,000 firms included in the 2018 Annual Business Survey used AI-related technologies, with the greatest adoption in the manufacturing and information sectors. Notably, average adoption was 18% when weighted by employment, as AI use skewed towards larger firms—more than 50% of firms with over 5,000 employees used AI to some degree. Similarly, Bonney et al. (2024) found that AI use was highest in the largest class of firms. Bonney and coauthors estimated that the AI use rate of American firms rose from 3.7% to 5.4% between September 2023 and February 2024, with an employment-weighted uptake rate of 20% over that period. In contrast to the lower overall adoption rates presented by the previous two studies, Roberts and Candi (2024) found that AI tools like ChatGPT were being applied to over half of the innovation projects of surveyed innovation managers in U.S. firms. Roberts and Candi also did not find a correlation between firm size and AI use, though they did find a negative correlation between AI use and firm age. This large spread in reported uptake was addressed by Baily et al. (2025), who compared firm uptake rates reported by the Census Bureau’s Business Trends and Outlook Survey and to a report from McKinsey & Company (~9% and 72%, respectively). The authors attribute the discrepancy to the size of the firms surveyed, noting that the Census Bureau’s survey was a representative sample of hundreds of thousands of American firms—few of which were large—while McKinsey conducted a convenience sample that overrepresented large corporations.

Nevertheless, other recent studies show that small- and medium-sized businesses have also begun to use AI tools at high rates. A 2025 report by the U.S. Chamber of Commerce found that 58% of surveyed small business said they used generative AI, rising from 40% in the previous year. The small businesses with the highest use rates were in the technology and financial services sectors (77% and 74%, respectively). Another recent survey conducted by the Initiative for a Competitive Inner City (ICIC) found that 89% of small business owners reported that at least one employee used AI tools. The most common type of employee reported as using AI was higher-level management and the most common use was data analysis.

In health care, Baten (2024) concluded that 18.7% of hospitals had adopted AI tools by 2022—primarily to predict demand for care and automate scheduling and workflow. Furthermore, Zink et al. (2024) found that Medicare billing volume for a specific AI-enabled clinical software used to diagnose heart disease grew more than 11 times between 2018 and 2022. On the other hand, only 17% of hospitals that conducted the type of scan used by the AI tool had adopted the AI-enabled technology. Diverging from the aforementioned studies, Nong et al. (2025) found that, by 2023, 65% of U.S. hospitals used predictive models—though less than half (44%) of the hospitals that used models evaluated them for bias, and only 61% evaluated them for accuracy.

Surveys have also revealed high levels of AI adoption in finance. A late 2024 survey from the accounting firm KPMG found that 71% of 2,900 companies around the world applied AI to corporate finance tasks to some degree, and 41% did so to a moderate or large degree. Of 300 U.S. companies included in the survey, 88% were using AI for corporate finance and 62% to a moderate degree or more. Both internationally and within the U.S., the most common finance tasks to which AI was applied were financial planning and accounting. A 2025 Deloitte Center for Controllership poll, meanwhile, found that over 80% of finance and accounting workers polled believed that AI tools would become standard for the profession within five years. That said, the poll also found that only 13.5% of the firms currently employing polled professionals were already using “agentic AI,” which refers to AI capable of making independent decisions to solve assigned tasks, a distinct tool from prompt-based generative AI tools like ChatGPT. The most commonly identified primary barrier to the implementation of agentic AI was a lack of trust in the technology, cited by 21.3% of respondents.

Government use of AI is also growing. Conducting a survey of the employees of unemployment insurance agencies, Rahman et al. (2024) found that 42.5% of unemployment agencies in the U.S. use AI as they attempt to prevent fraud. Similarly, a survey report by Ernst & Young LLP found that about half of public sector employees across all levels of government used AI applications at least several times a week.

Methodology

Our team obtained the survey data analyzed in this report by commissioning questions in the AmeriSpeak Omnibus survey instrument, which was fielded in the last week of June 2025. The Omnibus is a biweekly, multi-client survey of over 1,000 adults, mainly conducted online (roughly 90% of interviews) and to a lesser extent over the phone (about 10% of interviews). Omnibus participants are selected from the AmeriSpeak Panel using 48 sampling strata—including age, race, and education, and accounting for differences in population size and expected survey completion rates across strata—in order to form a representative sample of U.S. adults 18 years old and over.           

Operated by the National Opinion Research Center (NORC) at the University of Chicago, the complete AmeriSpeak Panel currently contains 65,884 panel members aged 13 and older across more than 58,000 households. Households are selected to join the panel from address-based sample (ABS) frames, primarily the NORC National Frame (61.8% of active AmeriSpeak households) but also including the United States Postal Service Delivery Sequence File (22.8%), a national consumer address file (9.2%), and voter registration files (6.2%). Once households have been identified, panel members are recruited via U.S. mail notifications, telephone interviews, and in-person field interviews.

Households added to the panel (meaning at least one member joins) have an initial recruitment rate of 24.4% over all recruitment years. The panel retention rate is 82%, meaning 82% of successfully recruited households are still active. Study-specific survey participation rates can apparently vary a great deal, from 20% to 70%.

The specific survey instance used in this report asked respondents to answer 14 questions, mostly concerning their use of AI tools. The survey had 1163 total respondents, though some of the questions targeted subsections of that population, such as small business owners (247 respondents) or people working in health care (147 respondents). Each response is weighed to account for each household’s likelihood of being selected for the sample and their demographic characteristics, as well as to adjust for nonresponses.

Finally, because this essay reports survey responses by individual users, survey results are an indication of respondents’ personal interpretation of their use of AI rather than an objective accounting of the prevalence of AI use. Respondents may have differing interpretations as to what constitutes AI usage, and some respondents may not be aware that they are using AI in certain applications.

Results¹

The survey’s questions fell within three broad classes. The first class of questions investigated respondents’ personal use and attitudes towards AI; the second class explored respondents’ use of AI in their professional lives; the third measured the use of AI in small businesses. Demographic information about respondents’ age group, sex, race and ethnicity, highest level of education, and annual household income was also recorded. Respondent ages were divided into ages 18–29, 30–44, 45–59, and 60+. The race and ethnicity groups were white non-Hispanic, Black non-Hispanic, Hispanic, and other or two or more non-Hispanic races/ethnicities. Education was broken down into no high school diploma, high school graduate or equivalent, some college or an associate’s degree, and bachelor’s degree or more.  The annual household income categories were less than $30,000 per year, $30,000 to less than $60,000 per year, $60,000 to less $100,000 dollars per year, and $100,000 per year or more.²

Personal use and attitudes toward AI

Among the full sample of 1,163 respondents, 57% report using generative AI for at least one personal purpose, most of whom use it for internet searches or web browsing (74%). The use of AI in a personal capacity is highest for more educated respondents. Sixty-seven percent of those with a bachelor’s degree or higher (BA+) use these tools in a personal capacity; 60% of those with some college or an associate’s degree (some college or an AA) use AI; and 46% of those with a high school diploma or equivalent use AI.

Frequency of personal AI engagement also scales with education. Fully 20% of BA+ respondents and 21% of respondents with some college or an AA engage with AI “daily or more,” versus 8% of high-school graduates and 8% of respondents without a diploma. Age-related differences in daily use parallel this trend but are not significant except for the contrast between 30–44 (18%) and 60+ (13%).

Around 40% of all respondents report that their use of AI has increased at least slightly compared to one year ago. In contrast, just 4% of respondents report a decrease in AI use. Notably, the 18–29 cohort is the age group with the highest proportion of respondents who say they use AI less frequently than they did a year ago, at 11%. This may be due to their higher AI use to begin with—only 26% of the 18–29 group say they did not use AI a year ago, compared to 30% of the 30–44 group, 40% of the 45–59 group, and 50% of respondents aged 60+. Changes in AI use over the past year also differ between education levels. Over half (55%) of respondents with a bachelor’s degree or higher (BA+) report some level of increased AI use in the past year. This is more than double the rate of increased use for respondents with no HS diploma (24%) and for those in the HS grad or equivalent group (27%).

Professional use of AI

Roughly one-in-five respondents (21%) report using generative AI in their professional role. This adoption rate follows a clear education gradient: 33% of those with a BA+ currently use these tools compared to 20% of respondents with some college or an AA, 12% of high-school graduates, and only 5% of individuals without a high-school diploma.

Age also matters: Usage peaks among 30–44 year-olds (31%) and remains high for those aged 45–59 (26%) and 18–29 (25%) but then plummets to 8% for adults aged 60 and above. AI use also varies across income levels, rising from 9% usage among earners below $30,000 to 34% among those making $100,000 or more. Men’s overall professional use (25%) slightly exceeds women’s (17%).

When it comes to institutional AI support on the job, usage is highest for document writing and editing—and again follows the education gradient. Among BA+ employees, 35% use AI for documents versus 16% of those with some college or an AA, 10% of high-school graduates, and only 2% of workers without a diploma. Men lead women in nearly every workplace-AI category except hiring and recruiting. Higher earners also report greater on-the-job AI adoption: 35% of respondents earning $100,000+ use AI for documents compared to 8% of those earning under $30,000.

Although a sizable minority (22%) of respondents say that over the last six months AI use in their workplace has increased, even more (61%) say that they are not sure or that the question is not applicable. Here, education drives the largest differences: 40% of BA+ respondents report increased use in the workplace compared with just 19% of those with some college or an AA, 9% of high-school graduates, and 5% of respondents without a diploma. Age contrasts are more modest, though the youngest cohort (18–29) experienced an increase in AI use in the workplace more frequently (24%) than the oldest (60+: 11%).

The impact of generative AI on worker productivity is often unclear, even to the workers themselves. Only 19% of all respondents report that AI increased their productivity in their daily tasks, and only 4% say it increased their productivity significantly. Even among respondents with a bachelor’s degree or more, just 28% say that AI increased their productivity in daily tasks. More than one in five respondents report that their daily productivity remained the same (22%) and over half of all respondents say they are either not sure about the effect of AI on their productivity or say it does not apply to them (53%). The rate of “Not applicable / Not sure” responses is highest among the 60+ age group at 72% and the no HS diploma group at 68%.

Finally, only 11% of all respondents anticipate that AI will increase job opportunities in their field over the next five years. Optimism is mildly higher among lower-income earners (<$30k: 15.6%) and high-school graduates (13%) than among BA+ holders (10%) and young adults (18–29: 10%).

A deeper dive into business use by sector and size

Within the subset of health care professionals, 53% of respondents report AI use. The application of AI cited by the greatest percentage of respondents is “patient communication tools” (25%). AI use is heavily skewed towards male health care professionals—82% of male health care professionals report AI use in their work, compared to only 40% of female health care professionals. Furthermore, high-income practitioners ($100k+ in household income) report AI use in their work at a rate of 77%, whereas lower-income staff report just 51% usage. Slightly complicating this story, the education level with the highest reported rate of AI use is the group with some college or an associate’s degree at 65%. This surpasses the use among BA+ respondents, which is only 60%.

Of the fewer than 50 respondents that work in the finance, insurance, or real estate industries, 62% said they use AI in their work. The most common use of AI by financial professionals is for customer service, identified by 35% of respondents. The 18–29 and 30–44 age groups have use rates of 77% and 89%, respectively, which are much higher than the 48% use amongst the 45-59 cohort and the 59% use amongst respondents aged 60+. As in the health care field, male financial professionals in our sample use AI much more frequently than their female counterparts, with 79% of men in finance reporting AI use compared to only 39% of women. However, in contrast with health care, the income group that uses AI most heavily in our finance sample is the less than $30k of annual income bracket.

Comparing AI use of respondents by firm size reveals remarkably similar trends across small and large businesses. Approximately 29% of sampled small businesses respondents use generative AI professionally compared to 27% of respondents employed at larger firms. Furthermore, 59% of small business respondents report that their workplace’s use of AI has increased over the last six months, while 60% of larger business respondents said the same. Our results could support the findings of earlier surveys which indicate that smaller firms have caught up to larger firms in AI adoption. The small business group contained a weighted base of 236 respondents, while the larger business group contained 627. Respondents who are not currently working are omitted from the comparison.

Though use rates are similar across small and large firms, our data indicate that there is a gap between personal and professional AI use rates. Furthermore, respondents seem skeptical about the impact of AI; over two thirds of respondents predict that AI will have no more than a slight impact on the number of jobs, and less than one in five respondents say that AI has improved their productivity even slightly. Finally, there remain significant demographic differences in AI use, including across age and levels of education or income.

Download the survey data