Microsoft Corp. says a powerful new artificial intelligence (AI) tool it has developed is capable of “medical superintelligence” and it has the experiment results to show it: The tool was able to diagnose disease four times more accurately and at a fraction of the cost of a panel of human physicians.
The software giant said its MAI Diagnostic Orchestrator (MAI-DxO) system queried leading AI models from OpenAI, Google, Anthropic, Meta Platforms Inc., and xAI that loosely mimics human experts working together. When it was done, it had outperformed human doctors, achieving an accuracy of 80% compared to the doctors’ 20%, based on 304 case studies sourced from the New England Journal of Medicine.
What is more, the software giant’s system also slashed costs by 20% by selecting less expensive tests and procedures — suggesting AI could help lower health care costs, a chronic issue in the U.S.
“This orchestration mechanism — multiple agents that work together in this chain-of-debate style — that’s what’s going to drive us closer to medical superintelligence,” Mustafa Suleyman, an ex-Google executive who is now CEO of Microsoft’s AI arm, told Wired.
The new Microsoft research achieved better results than previous work in the field because it paralleled how humans diagnose disease. It analyzed symptoms, ordered tests, and performed further analysis to reach its conclusions.
Microsoft hasn’t decided if it will commercialize the technology but it could be integrated into its Bing search engine to help users diagnose ailments. The company could also develop tools to help medical experts improve or automate patient care, a company official told the news organization.
Microsoft’s system is among the most promising in a fast-growing list of research projects in the last few years that show the potential of AI models and their ability to detect disease. Microsoft and Google have published papers proving that large language models (LLMs) can accurately pinpoint ailments when given access to medical records.
“This is an impressive report because it tackles highly complex cases for diagnosis,” said Eric Topol, a scientist at the Scripps Research Institute, adding that AI could, in theory, reduce the cost of medical care.
With so many fields primed for reimagination under the spell of AI, health care may hold the most promise, and impact, on society. The field has attracted $4.2 billion in global investments in AI startups in 2025, the most of any industry, according to Yijin Hardware, which analyzed sectors across key factors.
