Microsoft has developed an AI-powered medical tool called the “Microsoft AI Diagnostic Orchestrator” (MAI-DxO), which claims to diagnose complex medical conditions four times more accurately than human doctors. The system uses an “orchestrator” that manages virtual panels of AI agents acting like doctors, each responsible for different diagnostic tasks. Trained on 304 challenging medical cases from the New England Journal of Medicine, the AI achieved an 85.5% success rate, significantly higher than human doctors in the trial. The technology relies on large language models (LLMs) from OpenAI and other AI leaders, with OpenAI’s model performing best. Microsoft plans to integrate this AI into tools like its Copilot chatbot and Bing search engine, which already handle millions of health queries daily.
The AI health initiative is led by Mustafa Suleyman, who previously co-founded DeepMind, and is seen as a step toward “medical superintelligence” to help reduce staff shortages and long wait times in healthcare. Experts recognize the tool’s potential for improving diagnosis efficiency and reducing costs by minimizing unnecessary tests. However, the technology is still in early development, has not undergone peer review, and is not yet ready for clinical use. Microsoft has invested heavily in AI partnerships and emphasizes that the unique orchestration of multiple AI models is key to the system’s success. This breakthrough could transform healthcare by making diagnoses faster, cheaper, and more accurate.