Microsoft AI beats doctors in simulated diagnosis test

What happened: Microsoft said its MAI Diagnostic Orchestrator, an AI system using multiple agents, correctly identified 83% of complex medical cases in a simulated study published Oct. 22 in Science. Key numbers: AI 83% vs generalist physicians 36% with reference materials, 22% without. Emergency cases: AI about 80% vs 47%. 304 cases tested; physicians handled 56. Why it matters: The study was a simulation, physicians were generalists working alone, and authors work for Microsoft, so independent testing is needed before hospitals rely on it for patient care.

Microsoft said its MAI Diagnostic Orchestrator, an AI system using multiple agents, correctly identified 83% of complex medical cases in a simulated study published Oct. 22 in Science.

The study was a simulation, physicians were generalists working alone, and authors work for Microsoft, so independent testing is needed before hospitals rely on it for patient care.