Millions of Americans Are Talking to AI Instead of Going to the Doctor, and It’s Giving Them Horrendously Flawed Medical Advice

A JAMA Network Open study found 21 large language models performed poorly on realistic clinical cases, with especially high error rates in ambiguous scenarios.

A separate survey found that one in four U.S. adults has asked a chatbot for medical advice, and many have skipped seeing a clinician because of it.

The findings suggest people are increasingly using AI for health decisions even though current models can give unsafe or unreliable guidance.

Experts warn this could delay proper care and lead to worse outcomes when users trust chatbot advice over professional medical evaluation.