Harvard researchers find AI model outperforms emergency doctors in diagnosis

Researchers at Harvard Medical School have found that an artificial intelligence system can diagnose medical conditions better than emergency room doctors. The study compared an AI model created by OpenAI to physicians working with real patients in a Boston hospital. The AI successfully analyzed messy, complicated patient information and made better diagnostic decisions than the human doctors tested. One example involved correctly identifying lupus in a patient case. The research is important because it shows AI tools may work well in real emergency situations, not just in controlled laboratory tests. However, the findings raise new questions for hospitals. Doctors now face a challenge because patients increasingly arrive at appointments with AI-generated diagnoses from the internet. Medical professionals will need to decide how to use AI recommendations alongside information patients bring from online sources.