Microsoft's diagnostic AI reportedly scores 83% versus generalists' 36% in simulated complex cases

MAI-DxO coordinates several AI agents that act like a panel of doctors: They question a patient, order tests and debate a diagnosis. The reported scores come from simulated cases, not from treating patients.

The models behind the result

TechCrunch's reading of the study's data suggests that stronger underlying AI models account for much of the improvement, with Microsoft's coordination system adding gains on top. Bellan reports that GPT-5 scored 78.9% on complex cases on its own; DxO raised its accuracy by about four percentage points. In a different configuration, GPT-5.5 went from 72% alone to nearly 87% with DxO. Those figures compare AI configurations, not the system with physicians.

Microsoft AI CEO Mustafa Suleyman called the result a step toward “medical superintelligence.” TechCrunch reports that Microsoft has partnered with Beth Israel Deaconess Medical Center to evaluate DxO across hospital settings. Microsoft health vice president Dominic King described a plan in which consenting patients would receive standard treatment and engage with DxO in parallel. Its recommendations would then be compared with those of doctors.