More than 1 in 5 patients now use AI chatbots like ChatGPT, Claude and Gemini to research symptoms, interpret lab results and explore conditions, and that number is rising fast. In just four months, usage climbed from 18% to 21%, driven by patients using these tools more frequently, not just more broadly.
For provider organizations, this shift has real clinical implications. Patients arrive at appointments having already formed impressions about their health, shaped by tools that are easy to understand but not always trustworthy. Only 46% of patients consider chatbot health information accurate or reliable, yet nearly half never discuss what AI told them with their provider.
This gap between what patients read and what they share creates a silent variable in the exam room: patients may be acting on incomplete or incorrect information without their care team knowing.
Key findings relevant to provider organizations:
- 48% of AI health users don’t tell their doctor what the chatbot said, meaning the information is influencing behavior without clinical oversight
- Trust is highest for symptom lookup, lowest for ongoing condition management. Patients lean on AI to understand what’s wrong, but not how to manage it
- More than 1 in 5 patients using AI for health ask about mental health topics, and over 9 in 10 find the responses helpful, even as AI sycophancy poses documented risks in this category
Understanding how your patients use AI before and between visits is an important step toward closing the information gap and strengthening the patient–provider relationship.
