Key takeaways
- In a U.S. survey of adults, the share who had used AI chatbots to look up health information rose over 1 year from 16% to 32%.
- Among users, 81% said they had taken some action after seeing an answer, and 40% said they had consulted a healthcare professional, such as a doctor or nurse.
- Among AI users, the share who said they trusted health information provided by healthcare professionals was 85%. Rather than replacing hospitals, AI may become a place people turn to for information before an appointment.
Survey findings - Who asks AI, and what do they do next?
Rock Health, a digital health investment and advisory firm, surveyed 8,000 U.S. adults in December 2025 and published the findings in March 2026.
- Use over time (same annual survey): 2024: 16% → 2025: 32% (one or more of symptom-based diagnosis, treatment, medication, or other health information)
- By generation: Millennials 48% · Gen Z 45% · Gen X 25% · Baby Boomers 12% · Silent Generation 7%
- Services used (all respondents): ChatGPT 23% · Healthcare-provider chatbots 5%
- Topics of use (users): Treatment options 59% · Symptom-based diagnosis 56% · Medication and side effects 55%
- Actions afterward (users): Additional search 42% · Consulted a healthcare professional 40% · Adjusted medication 18%
- Trust in health information (among users, completely or somewhat): Healthcare professionals 85% · AI chatbots 56%
Academic research offers another data point. A paper published in JMIR in 2024 found that 21.5% of 2,406 U.S. adult respondents surveyed in 2023 had used ChatGPT to find health information. However, the survey drew on registered research participants (with an 11.2% response rate), so it is difficult to generalize the findings to all U.S. adults or compare them directly with Rock Health’s figures.
The Institute’s interpretation - How should we read the “doubled” figure?
This is Rock Health’s own survey, based on U.S. data; it does not address patients or hospital choices in Korea. The implications for hospitals are the Institute’s interpretation.
A domestic survey also found that 48.9% had experience seeking health advice from AI, while 69.3% said they intended to use it in the future. Preparing for patients who come in after reading AI answers is already a practical issue for hospitals in Korea.
Change is moving quickly, but patients have not left healthcare professionals behind. Hospitals need to be ready to meet patients who come in after reading AI answers.
- Generational differences are substantial. Millennials: 48%; Baby Boomers: 12%. Depending on a hospital’s main patient group, this shift may become noticeable at different times.
- AI answers may carry into the consultation room. 40% of users said they consulted a healthcare professional, and 18% said they adjusted their medication. If a similar pattern emerges in Korea, checking and clarifying what patients have read may become part of a consultation.
- Use is growing quickly, but building a reliable information base takes time.Clinician-reviewed care information is difficult to build up quickly. It may be worth beginning review and organization before demand grows further.
Are you ready to meet patients who have seen AI answers?
- Can consultation records distinguish cases where patients mention AI answers?
- Have clinicians reviewed guidance on treatment options and medication use?
- Have you decided which care information to organize first based on the ages of your main patient groups?
The “doubled” figure describes a 1-year change in the U.S., not a forecast for Korea. Still, a single clinician-reviewed article about treatment guidance can remain a useful asset for any hospital.
Sources
Rock Health, analysis related to “2025 Consumer Adoption of Digital Health Survey” (published 2026.3.23; self-conducted)
Original source: Rock Health
Paper: Ayo-Ajibola et al., Journal of Medical Internet Research (2024.8.14) JMIR · PMC
Korea Press Foundation <미디어 서베이> 2026 Issue 1: Foundation materials · 기자협회보 · Yonhap News Agency (2026.4.15)