Microsoft says it has developed an AI system that creates a ‘path to medical superintelligence’ that can deal with ‘diagnostically complex and intellectually demanding’ cases and diagnose disease four times more accurately than a panel of human doctors.
[ https://microsoft.ai/wp-content/uploads/2025/06/MAI-Dx-Orche…0x1498.jpg https://microsoft.ai/new/the-path-to-medical-superintelligence/
[ https://arxiv.org/abs/2506.22405](https://arxiv.org/abs/2506.
“Benchmarked against real-world case records published each week in the New England Journal of Medicine, we show that the Microsoft AI Diagnostic Orchestrator (MAI-DxO) correctly diagnoses up to 85% of NEJM case proceedings, a rate more than four times higher than a group of experienced physicians. MAI-DxO also gets to the correct diagnosis more cost-effectively than physicians.”
AI that thinks like a doctor: a new era in medical diagnosis.
Imagine walking into a doctor’s office with a strange set of symptoms. Rather than jumping to conclusions, the doctor carefully asks questions, orders tests, and adjusts their thinking at every step based on what they learn. This back-and-forth process—called sequential diagnosis—is what real-world medicine is all about. But most AI systems haven’t been tested this way. Until now.
A new benchmark called Sequential Diagnosis is flipping the script.
Developed using over 300 complex cases from the New England Journal of Medicine, it challenges both doctors and AI systems to solve medical mysteries step by step—just like a real clinical encounter. The AI isn’t spoon-fed all the data. Instead, it has to ask for the right information, interpret results, and refine its hypotheses in real-time.
To tackle this, researchers created MAI-DxO, an AI “orchestrator” that behaves like a smart diagnostic team. Think of it as a conductor coordinating a group of virtual physicians. It proposes potential diagnoses, picks high-value tests (without wasting money), and adjusts its decisions as new data comes in. When paired with OpenAI’s o3 model, MAI-DxO achieves 80% diagnostic accuracy—a striking leap from the 20% baseline of human generalist doctors—and does so while cutting diagnostic costs by 20% compared to humans, and 70% compared to standard AI.
When tuned for maximum performance, it even hits 85.5% accuracy.
What’s even more exciting? This system works not just with OpenAI’s models, but also with Gemini, Claude, Grok, DeepSeek, and Llama—suggesting a robust, model-agnostic approach to reasoning.
According to Microsoft, this may be a critical step toward what they call “medical superintelligence”—a future where AI doesn’t just memorize textbooks but thinks like a physician, delivering more accurate, timely, and affordable care.
There are caveats and other use cases aside from diagnosis but this is early stage.
We make responsible AI to empower people’s lives.