AI Doctors Beat Humans In Study

For patients facing difficult diagnoses, the future of medical care may look less like a single doctor’s intuition and more like a coordinated digital team. A study published this week in Science reports that Microsoft’s new AI diagnostic system, known as MAI-DxO, correctly identified complex medical conditions in a simulation more than twice as often as human generalist physicians. The result marks a significant shift in how technology companies are positioning their tools to assist clinicians in the Great Lakes and Ohio Valley regions where healthcare demand is high.

The system operates by coordinating multiple AI agents that function similarly to a panel of doctors. These agents question patients, order tests, and debate potential diagnoses before reaching a conclusion. In tests using complex cases from the New England Journal of Medicine, the AI achieved an accuracy rate of 83 percent. By comparison, primary care doctors and general internists working individually with reference materials were correct only 36 percent of the time. It is important to note that such complex cases are typically handled by specialists or teams in real settings, who were not included in this specific comparison.

Microsoft AI CEO Mustafa Suleyman described the findings as a step toward what he calls medical superintelligence. He noted that the company is currently collaborating with the Mayo Clinic to train a foundation model for healthcare, with a release planned for next year. Suleyman believes this infrastructure will allow for enormous leaps in clinical decision support. He predicts that by 2027, the technology will enable highly accurate predictions about how medical records might unfold, a capability he describes as gold dust for clinicians.

However, the data suggests that much of the performance improvement comes from using powerful underlying models rather than the orchestration system itself. Microsoft tested the system with models from various companies, finding the best results with OpenAI’s GPT-5 and GPT-5.5. While the orchestration added some gains, the gap between weaker and stronger base models remained larger than the boost provided by the AI panel. For instance, a weaker model saw its accuracy rise from 46.4 percent to nearly 69 percent with the system, but stronger models saw smaller relative improvements.

Cost analysis also revealed mixed results. Running the system with GPT-5 alone cost about $8,400 per complex case with 78.9 percent accuracy. The orchestrated system reduced the cost by approximately $3,000 but increased accuracy by only about four points. On emergency department cases, there was little improvement in either accuracy or cost when comparing the base model to the orchestrated version. This suggests the technology is currently most effective for complex, non-urgent scenarios where thorough deliberation is possible.

To test these capabilities in real-world settings, Microsoft has partnered with Beth Israel Deaconess Medical Center in Boston. The study will evaluate the AI across various hospital environments, including emergency departments and urgent care. Patients will receive standard treatment while also engaging with the AI system if they consent. Researchers will then compare the AI’s recommendations against those of human doctors. Dominic King, vice president of Health at Microsoft AI, called the current mismatch between medical expertise supply and demand the biggest challenge in healthcare delivery, and this partnership aims to address that gap.

This move places Microsoft in direct competition with Google and OpenAI, both of which have recently claimed their AI systems can match or outperform doctors. Microsoft has defined medical superintelligence as the capacity to surpass contemporary human-led health systems in reasoning and coordination across the full arc of care. The company aims to integrate these tools directly into clinical workflows, moving beyond simple benchmarks to improve patient outcomes. As Suleyman stated, the company is committed to accelerating the introduction of this technology to the world.