One in Three American Adults Now Uses AI Chatbots for Health Information

The doctor’s waiting room has a new competitor, and it doesn’t require an appointment, a co-pay, or even pants. On August 25, 2026, the Pew Research Center published a survey showing that 34% of American adults — roughly one in three — now use AI chatbots for at least one health-related reason. The report, titled “From Diagnoses to Treatments, Why Americans Use AI Chatbots for Health,” is based on a survey of 3,488 U.S. adults conducted June 22–28, 2026, and it paints a picture of a population that has quietly folded conversational AI into one of the most sensitive corners of daily life.

What people are actually asking chatbots

The headline number conceals a remarkably practical mix of use cases. The single most common motivation is speed: 28% of U.S. adults say they use chatbots to get health information quickly. Close behind, 25% use them to figure out what’s causing symptoms — effectively self-diagnosis before (or instead of) a doctor’s visit.

Cost matters too. 22% say they turn to chatbots because the information is low-cost or free, a figure that doubles as an indictment of American healthcare economics. The same share (22%) use chatbots to learn about treatments or to better understand a diagnosis a doctor has already given, and 20% use them to interpret lab results — the classic “your results are online, good luck decoding them” scenario that has sent patients to search engines for two decades.

Two motivations stand out for what they reveal about human behavior. 18% of adults use chatbots to get health information they’re uncomfortable talking about with another person — the embarrassing symptoms, the awkward questions. And 15% use them to help decide whether they need to see a doctor at all, making the chatbot a triage layer sitting in front of the healthcare system.

Nearly all chatbot health users cite multiple reasons. This isn’t a novelty; for a third of the country, it’s infrastructure.

Who’s using them

The demographic splits are sharp. Asian Americans (56%) are the most likely to be chatbot health users, followed by Hispanic adults (39%) and Black adults (35%), with White adults at 29% — though Pew notes the Asian estimate covers English speakers only.

Age is the steepest gradient: 44% of adults 18–29 and 40% of those 30–49 are chatbot health users, versus just 17% of those 65 and older — the group that arguably uses the healthcare system the most. Education and income track the same direction: college graduates (44%) and upper-income adults (48%) far outpace those with a high school education or less (23%) and lower-income adults (28%).

Attitude toward AI is the strongest predictor of all. Among Americans who say increased AI use makes them more excited, 60% are chatbot health users. Among those who feel more concerned, only 21% are. In other words, the people using chatbots for health are largely the people who already liked AI — for now.

Helpful? Overwhelmingly. Trustworthy with your data? Complicated.

Here’s where the survey gets interesting. Among chatbot health users, 47% say the health information they get is extremely or very helpful, and another 48% say it’s somewhat helpful. Only 5% found it not too or not at all helpful. That is an extraordinarily lopsided satisfaction rating for any consumer technology, let alone one operating in a domain where errors carry real consequences.

But satisfaction with the answers doesn’t translate into comfort handing over the inputs. When asked about sharing personal health information with a chatbot, users split almost down the middle: 29% say they’d be extremely or very comfortable, 42% somewhat comfortable, and 26% not too or not at all comfortable. Pew contextualizes this unease by noting that some platforms have already faced scrutiny for storing and selling health data, or for using it to train large language models — even as tech companies race to unveil dedicated healthcare chatbot products.

The contrast is the story: people find the output valuable enough to keep using, while remaining wary about what happens to the details they type in.

The mental health caveat

A companion Pew report published the same day adds a sobering counterweight. Asked whether chatbots help or hurt people using them for emotional struggles, Americans come down clearly on the “hurt” side: 39% say chatbots do more to hurt people using them for loneliness (vs. 19% who say help), and 36% say the same about depression (vs. 17% help). The pattern extends to AI’s broader societal effect, and it aligns with a growing research literature linking chatbot-based social support to increased, not decreased, loneliness.

So the public’s position is more nuanced than either boosters or doomsayers would like: chatbots are trusted informational tools for decoding symptoms and lab results, but suspected liabilities as emotional crutches.

Why this matters

Healthcare has always been the highest-stakes destination for consumer AI — high error costs, deep information asymmetry, and heavy regulation. The Pew data shows the adoption curve has already rounded the corner without waiting for permission. When a quarter of adults are using a chatbot to decide whether symptoms warrant a doctor, the chatbot is no longer an accessory to the healthcare system; it’s a gatekeeper in front of it.

That raises urgent questions the survey doesn’t answer: Are chatbot answers accurate enough for triage? What happens to the symptom data users volunteer? Will regulators treat health-flavored general chatbots with the seriousness of medical devices? And what does it mean for health equity that usage skews young, educated, and affluent — while the heaviest healthcare consumers, seniors, largely stay away?

What’s clear is that the behavioral shift is done. The average American has already decided that asking an AI about their rash at 2 a.m. is normal. The governance, safety standards, and privacy guarantees are now racing to catch up.