JAMA Medical News AI Tools for Patients Have Arrived
5 snips
Mar 6, 2026 Rita Rubin, lead senior staff writer for JAMA Medical News who covers clinical medicine and health policy, walks through the rise of patient-facing AI. She discusses ChatGPT Health, what these tools can and cannot do, triage and accuracy concerns, privacy and HIPAA gaps, and government plans to build autonomous AI for conditions like heart failure.
AI Snips
Chapters
Transcript
Episode notes
Generative AI Is Becoming Patients' First Stop
- ChatGPT Health and other patient-facing generative AI are rapidly replacing Dr. Google for health queries.
- OpenAI reports 800 million weekly ChatGPT users and launched ChatGPT Health after one in four users asked health questions.
Prefer Tools Trained On Verified Medical Data
- Recognize that not all patient-facing AI train exclusively on verified medical sources; vet tools accordingly.
- Some emerging apps train only on verified medical information, unlike widely used chatbots with mixed training data.
Chatbot Triage Is Least Reliable At Extremes
- Accuracy of chatbot health advice is uneven and can be problematic for serious or trivial conditions.
- A study using physician-authored vignettes found the chatbot's triage recommendations were least accurate for the most and least serious cases.