AI Chatbots Are Bad at Diagnosing Symptoms For a Surprising Reason, Study Finds

A study tested how well large language model chatbots help people handle common health problems and found users using chatbots identified conditions less accurately than controls.

When scenarios were given directly to models without human users, chatbots identified relevant conditions and often suggested appropriate care levels.

The main issue was communication breakdowns in human–chatbot interactions, not lack of medical knowledge.

Researchers advise policymakers to assess real-world performance before deploying AI in frontline healthcare.