Microsoft Says Its New AI Is More Than Twice as Good at Diagnosing Diseases as Doctors
Key Points:
- Microsoft’s AI Diagnostic Orchestrator (MAI-DxO) demonstrated more than twice the diagnostic accuracy of practicing generalist physicians in complex medical cases, achieving 83.2% accuracy compared to 36.1% for doctors with reference materials.
- The AI system operates sequentially, mimicking real-world diagnostic processes by asking follow-up questions and ordering tests before reaching a diagnosis, which helps reduce unnecessary procedures and costs.
- MAI-DxO’s performance was validated in peer-reviewed research published in Science, with plans for clinical testing at Beth Israel Deaconess Medical Center in Boston set to begin soon.
- The AI’s diagnostic accuracy was strongest when using OpenAI’s GPT-5 and improved further with GPT-5.5, and it is designed to mitigate cognitive biases common in human physicians by simulating a panel of diagnostic roles.
- While MAI-DxO outperforms generalists, researchers expect the accuracy gap to narrow when compared to specialist doctors, and cost savings are currently based on simulated estimates rather than real-world expenses.