Microsoft AI Model Outperforms Human Doctors in Medical Diagnosis Benchmark

Microsoft researchers published a study in the journal Science demonstrating a new artificial intelligence system that diagnoses medical cases sequentially. The tool mimics the step-by-step methodology of human doctors by asking targeted questions, ordering specific tests, and iteratively refining its clinical hypotheses.

The orchestrator layer achieved an outstanding 83.2% accuracy rate when evaluating complex medical cases. In comparison, human generalist physicians with access to reference materials averaged just 36.1% accuracy, marking a major milestone toward achieving highly specialized medical superintelligence.