Microsoft AI CEO Mustafa Suleyman told The Verge, “We’re going to build something that is indistinguishable from human performance … in the next couple of years.” He viewed healthcare as a clear area where AI can now improve lives.
The company asserts it has built an AI tool capable of diagnosing conditions four times more accurately than human doctors. Published Thursday in *Science*, the findings compare the system against human physicians using scenarios from medical case reports. Microsoft compiled these into an open-source benchmark called SDBench.
The Diagnostic Orchestrator uses multiple AI agents to mimic a panel of experts, reasoning through symptoms and deciding on tests. It achieved 83 percent accuracy on cases from *The New England Journal of Medicine*, outperforming generalist physicians (36 percent) and those without references (22 percent). In emergency department cases at Beth Israel Deaconess Medical Center, the system scored 80 percent compared to 47 percent for human doctors, while also reaching diagnoses at a lower estimated cost.
Powered by OpenAI’s GPT-5, the system outperformed other frontier models acting alone, though newer models like GPT-6 Astra were not included. Microsoft notes that improvements in model capability do not diminish the benefits of the multi-agent approach.
Microsoft admits the technology is not ready for widespread deployment. While the system mirrors the diagnostic process, it has only been tested on past cases. Suleyman emphasized the need for “clinical validation” to move forward.
This research supports Microsoft’s “medical superintelligence” goal. AI health lead Dr. Dominic King stated the company will begin testing at Beth Israel Deaconess in the next couple of months. King stressed the importance of a responsible rollout due to regulatory compliance, though commercialization timelines are pending.
Suleyman predicts “extremely impressive deployments” of intelligent systems next year that will transform care delivery. He maintains a “human-centered” vision where AI removes “drudgery,” allowing clinicians to spend more time with patients.
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.