United States: 72% of Patients Want to Be Informed About AI Usage

Nearly half of Americans are unaware if AI has already played a role in their care, and only a minority feel very comfortable sharing their data with chatbots. At the same time, a large majority demands that providers disclose any use of AI, particularly for medical decisions. The Pew Research Center details that established practices exist on both the patient and physician sides, but trust remains cautious.

Lack of Awareness and Reported Comfort with AI Tools

47% of Americans do not know if AI has ever played a role in their treatment. Opinions are divided on sharing medical history and diagnoses with AI systems: 29% say they are extremely or very comfortable sharing health information with chatbots, while 26% report they are not. Overall, respondents find these tools convenient but not entirely trustworthy.

Disclosing AI Use Becomes a Majority Expectation

72% of respondents believe it is extremely or very important for their medical provider to disclose any use of AI. Eight in ten want to be informed when AI is directly involved in care or medical decisions, such as for a diagnosis or the analysis of tests. Transparency is thus deemed crucial by patients, and concerns appear even more pronounced when AI is used in clinical settings.

On the Patient Side: Reasons for Use and Perceived Utility of Chatbots

A quarter of respondents report using chatbots to diagnose symptoms, and recent figures indicate that about one-third have consulted them for health-related questions. The primary reasons cited are convenience and cost, with 24/7 availability and often free access, sometimes without the need for an account. Mentioned uses include quickly searching for information, understanding a diagnosis, interpreting lab results, and helping decide whether to seek in-person consultation. Among users, 47% find the information "extremely" useful and 48% "fairly" useful, compared to 5% who deem it not useful or of little use. Chatbots simulate conversation and respond based on the data on which their models have been trained.

In Clinical Settings: Note-taking, Triage, and Reduced Exam Wait Times

Generative AI is also making its way into clinics, where it is used for note-taking and interpreting exams. It is frequently employed for triage and to expedite certain clinical tasks. One reported example noted a reduction in the interpretation time for X-rays from 11.2 days to 2.7 days. Reported results also indicate performance superior to that of human emergency physicians for triage.

Survey Methodology and Known Risks of Chatbots

The Pew Research Center surveyed 3,488 American adults between June 22 and June 28. In a context where chatbots and AI are increasingly present both in and out of clinics, transparency remains a major concern. The limitations of chatbots are also highlighted: they can generate false information, as their responses prioritize probability over truth and often appear very confident, which can be problematic in healthcare. The results suggest that public trust and understanding are not evolving at the same pace as adoption.