Roughly half of American adults now use AI chatbots, but the other half has stayed out, and the reasons they give are more about trust and interest than about technical know-how. This article looks at Pew Research Center’s 2026 findings on who is opting out and why.
How many Americans actually use chatbots?
According to Pew Research Center’s “Americans and AI 2026” report, published June 17, 2026, 49% of US adults say they use AI chatbots, up from 33% in 2024. The survey covered 5,119 adults and was fielded February 17-23, 2026, through Pew’s American Trends Panel.
That leaves 51% who never use them. Pew describes this as about half of adults. Put another way, the two groups are almost exactly the same size, which is a notable change for a product category that was a novelty only a few years ago.
Use is also not casual for everyone who has tried it. Pew found that 24% of adults use chatbots daily, made up of 8% who use one about once a day, 12% who use one several times a day and 4% who use one almost constantly. Another 25% use them several times a week or less.
A simple check of those groups: 24% daily plus 25% less often gives 49%, which matches the overall user share.
Who are the people who never use them?
Age is the sharpest dividing line. Pew’s figures for the share of each age group that never uses chatbots:
| Age group | Never use chatbots |
|---|---|
| 18-29 | 34% |
| 30-49 | 39% |
| 50-64 | 58% |
| 65 and older | 77% |
The gap between the youngest and oldest groups is 43 percentage points (77% minus 34%). Pew summarizes the oldest group bluntly: roughly three-quarters of adults 65 and older say they never use chatbots.
Even among the youngest adults, though, about a third are non-users. Opting out is not only an older-generation story.
Pew also reported differences by race and ethnicity. Among adults who never use chatbots, the shares were 54% of White adults, 50% of Hispanic adults, 51% of Black adults and 29% of Asian adults (English speakers only for the last group). Those are descriptive splits from one survey, and they do not on their own explain why the patterns differ.
Why do non-users say they stay away?
Pew asked non-users to rate several possible reasons as a major reason, a minor reason, or not a reason. The results, among adults who never use chatbots:
| Reason | Major reason | Major or minor |
|---|---|---|
| Not interested in using them | 60% | 83% |
| Concerned about how personal information will be used | 54% | 79% |
| Don’t trust them to give accurate information | 45% | 76% |
| Don’t know how to use them | 29% | 55% |
| Think others will judge them for it | 3% | 14% |
Several things stand out. Lack of interest leads, which suggests that many non-users do not feel they are missing anything. Privacy and accuracy are close behind, and each is named as at least a minor reason by roughly three in four non-users. Social stigma barely registers, with only 3% calling it a major reason.
The percentages are not meant to add to 100. Each person could rate every reason, so a single non-user can say they are uninterested, worried about privacy and unsure of accuracy all at once.
Is it a skills problem or a trust problem?
The data lean toward trust and interest. Not knowing how to use chatbots was a major reason for 29% of non-users, compared with 54% for privacy and 45% for accuracy. Still, 55% named lack of know-how as at least a minor reason, so a learning barrier is real for many people.
That matters for anyone building or explaining these tools. A friendlier tutorial may help the group that does not know where to start, but it will not by itself answer worries about personal information or wrong answers.
The accuracy worry is grounded in how these tools work. Our explainer on what generative AI tools can and can’t do for work covers why a fluent answer is not always a correct one, which is the issue non-users describe when they say they do not trust the information.
Will non-users change their minds?
Not soon, if their own forecasts hold. Pew asked non-users how likely they were to try a chatbot in the next 12 months:
- 67% said not too or not at all likely, including 40% who said not at all likely.
- 16% said somewhat likely.
- 5% said extremely or very likely.
- 12% were not sure.
Those four shares sum to 100% (67 + 16 + 5 + 12), so they cover every non-user. Only about one in twenty expects to become a heavy adopter within a year.
Adoption has still grown quickly. The rise from 33% in 2024 to 49% in 2026 happened even though many people said they were not interested. Intentions captured in a survey are a snapshot, and tools built into phones, search engines and workplace software can pull people in without a deliberate decision to “start using a chatbot.” Pew’s data do not measure that effect, so it should be treated as a possibility rather than a finding.
What do users and non-users think about AI’s pace?
Opting out of the tools does not mean being unaffected by them, and it does not mean ignoring the debate. In the same Pew survey, 63% of adults said AI is advancing too quickly, 19% said about the right pace, 16% were not sure and 2% said too slowly.
Views about impact are mixed. Looking 20 years ahead, 40% of adults expect AI’s effect on society to be negative and 16% expect it to be positive, with 31% saying equally positive and negative and 13% unsure. For their own lives, 31% expect a negative effect and 23% a positive one, while 27% say equally both and 19% are unsure.
Younger adults are not simply cheerleaders. Negative views of AI’s impact on society were most common among those 18 to 29, at 48%, compared with 39% of those 30 to 49 and 37% of those 50 and older. Heavy use and worry can sit side by side, a tension we explored in our opinion piece on how AI chatbot use is outpacing trust.
What can readers take from the numbers?
For people who do not use chatbots, the survey is a reminder that the choice is common. Half of adults are in the same position, and no stigma attaches to it, since only 14% say they think others would judge them at all.
For people who do use them, the privacy and accuracy concerns are worth taking seriously rather than dismissing. A few habits follow directly from the reasons non-users cite:
- Avoid entering sensitive personal details unless you understand how a service stores and uses them.
- Check important answers, such as medical, legal or financial ones, against a primary source.
- Treat a confident tone as a style, not as evidence of accuracy.
- Review the privacy settings of any assistant built into a device or app.
These steps also apply to the newer wave of software that acts on a user’s behalf, which we describe in what AI agents can actually do today. The more a tool is allowed to do, the more the privacy and accuracy questions matter.
Reading the data with care
A few limits are worth keeping in mind. The figures are survey self-reports from February 2026 and may have moved since. Pew defines chatbot use in its own way, so people who interact with AI features they do not think of as chatbots may be counted as non-users. And the survey sorts people by whether they use chatbots, not by how much they know about them.
The upshot is that the market is split almost evenly, and the divide is not mainly about ability. Lack of interest, privacy and accuracy lead the list of reasons, and the stated reluctance to adopt is firm. Any company or employer that assumes everyone will eventually come around should look at those numbers first.
A divided market with clear reasons
Chatbots have reached about half of American adults faster than most consumer technologies, yet the other half has specific, repeatable reasons for holding back. Interest, privacy and accuracy matter more than confusion or embarrassment, and few non-users expect to change course within a year. Understanding those reasons is a better guide to where the technology goes next than the headline adoption figure alone.
This article is general information about survey findings, not advice on whether to use any particular product. Survey results reflect the dates and methods Pew describes.
Sources: Pew Research Center
