(Credit: © ChayTee - stock.adobe.com)
Half of Survey Respondents Would Accept a Reassuring Chatbot Answer Without Question
In a Nutshell
- Among Americans who use AI for health, 60% said they would believe AI-generated health advice even if it contradicted guidance from a healthcare provider.
- 63% of AI users said they had followed AI health advice without confirming it with a provider, and 58% of millennial parents who use AI for health said they had acted on its advice for their children without consulting one.
- Many users shape what the AI sees: 53% have reworded a question after disliking the answer, and 42% admitted lying about who a health question was about for privacy reasons.
When a doctor and a chatbot disagree, plenty of AI users are ready to side with the chatbot. In a new survey of 2,000 adults, 60% of respondents who use AI for health purposes said they would believe AI-generated health advice even if it contradicted guidance from a healthcare provider.
That is heavy faith to place in software that has never taken a pulse or read a chart nobody gave it. Talker Research, which ran the poll, reports that one in seven Americans trust AI chatbots as much as they trust a doctor. Among people who already lean on the tools for health questions, that confidence can stretch all the way to overruling a professional.
And the trust is not staying hypothetical. In the same poll, commissioned by virtual care company MD Live by Evernorth, 63% of AI users said they had followed health advice from AI without confirming it with a healthcare provider.
Survey of 2,000 Americans on AI health advice
Talker Research surveyed two groups between June 18 and July 6, 2026. One was made up of 1,000 millennial parents, split evenly between men and women. A second group included 1,000 Americans from the general population. Half of each group used AI for health purposes, giving the pollsters a large pool of people who already bring medical questions to chatbots.
Like any poll, the results reflect what people reported about their own habits and beliefs. Nobody’s chatbot conversations were observed, and nothing in the published results describes checking whether the advice people received was accurate.
Among AI health users, the habit runs deep. They ask an average of nine health questions a week, which works out to more than one a day. When asked why they turn to AI, 37% pointed to convenience, 31% said it was faster than seeing a doctor and 30% said they wanted reassurance.
Reassurance deserves a closer look. People seeking it are hoping to hear a particular kind of answer, and several of the survey’s other findings point to users steering toward exactly that.
Acting on AI health advice without a second opinion
Half of respondents said they are likely to accept an AI answer without questioning it if the response is clear or reassuring. A polished, confident paragraph, in other words, can be enough to settle the matter.
Double-checking is far from routine. Some 28% of respondents said they verify the answers they get from AI less than half the time.
Some learned too late that the advice was a poor fit. In the poll, 59% of AI users said they later realized health information from AI didn’t fully apply to their situation, and 49% said they had acted on that information before discovering it wasn’t the right fit. A general answer about a symptom or condition can miss the detail that makes one person’s case different, such as another medication they take or a condition they didn’t mention.
Wording plays a bigger role than most users seem to realize. Roughly eight in 10 (78%) didn’t know that the way a question is phrased affects the answer an AI gives. Meanwhile, 53% admitted to rewording a question after they didn’t like the response they got.
Dr. Maggie Williams, medical director of virtual primary care at MD Live by Evernorth, sees a pattern in those numbers. “The survey indicates that people are often going to AI already hoping for a certain answer,” she said.
Many users also hold mistaken ideas about where the answers come from. Just over one in five (22%) believe AI only draws from trusted medical sources. A similar share of regular users believe the AI factors in their personal medical history, even when they haven’t provided it. If that history has not been shared with the AI or made available to it, the tool may not know about a past surgery or a current prescription, so someone who assumes otherwise may be trusting advice built on missing information.
Parents, privacy and the questions behind the answers
Parents are leaning on the technology for their children, too. Among millennial parents who use AI for health, 58% said they have acted on AI health advice for their kids without consulting a provider.
Children’s health questions can involve details such as age, weight and symptoms, which may be missing from a vague AI prompt.
Privacy worries add another wrinkle. In the survey, 42% admitted to lying about who a health question was about for privacy reasons. Someone might frame a question about their own symptoms as being about a friend, for example. That kind of fib may protect personal information, but it also means the AI is working from an inaccurate picture of the person who actually needs the answer.
A survey with a stake in the answer
Users are not blind to the risks. Their top concerns about AI health advice were getting incorrect information (34%) and the lack of human oversight (31%). And 27% said they would trust AI more if a doctor reviewed it.
Such a finding lines up neatly with the business of the company that paid for the poll. MD Live by Evernorth provides virtual medical care, so a survey showing that people need clinicians to check their chatbots supports its own pitch. Readers should keep that commercial interest in mind, even though the figures still offer a snapshot of how respondents say they use and view AI for health questions.
Williams expects chatbot use to keep growing regardless. “The results show people will continue to turn to AI for their health care questions,” she said. In her view, the value of a physician lies in asking follow-up questions and placing information in the context of a patient’s health history. As she put it, “healthcare providers add important clinical guidance by catching details a patient may not consider relevant.”
A chatbot can only work with the story it is given, and this survey shows many users are handing it an edited version: a reworded question, a missing medical history, sometimes a made-up patient. Believing that answer over the advice of someone who has actually examined the patient puts a lot of weight on a very thin foundation.
Survey Notes
Methodology
Talker Research conducted the survey between June 18 and July 6, 2026, polling 1,000 millennial parents (evenly split by gender) and 1,000 Americans from the general population. Half of each group used AI for health purposes. Results are based on self-reported answers, which can be affected by memory, how questions were worded, and a desire to give socially acceptable responses. Talker says the survey was administered and conducted online, and it describes the general-population AI users as having internet access. It does not report a margin of error or give recruitment or weighting details. It also does not specify the exact base for the “one in seven Americans” figure, so that number is presented here as Talker reported it.
Funding and Disclosures
MD Live by Evernorth, a virtual care company, commissioned the survey. Its medical director of virtual primary care, Dr. Maggie Williams, provided the expert commentary in the source article. Because the company provides clinician-led virtual care, it has a commercial interest in findings that point to the value of physician oversight.
