Microsoft has announced a significant leap toward what it calls "medical superintelligence," unveiling a cutting-edge AI system capable of diagnosing diseases with four times the accuracy and at a fraction of the cost of a team of human doctors.
The new tool, known as the MAI Diagnostic Orchestrator (MAI-DxO), was tested using 304 complex case studies from the New England Journal of Medicine. It operates on a benchmark designed to simulate the step-by-step diagnostic reasoning typically followed by human physicians. The AI draws on several leading language models—including OpenAI’s GPT, Google’s Gemini, Anthropic’s Claude, Meta’s Llama, and xAI’s Grok—working in tandem to replicate the collaborative process of a team of medical experts.
In Microsoft's evaluation, the AI system achieved an accuracy rate of 80–85.5%, while experienced physicians averaged just 20%. It also reduced diagnostic costs by 20% by recommending more affordable tests and procedures. Mustafa Suleyman, CEO of Microsoft's AI division, called the breakthrough a "big step towards medical superintelligence," highlighting the system’s unique chain-of-debate approach as key to its performance.
While AI is already playing a role in healthcare—for example, assisting radiologists with scan interpretation—Microsoft’s system aims to more closely mirror how doctors approach complex diagnoses by ordering tests, asking targeted questions, and refining their assessments over time. The company stresses that this technology is designed to support, not replace, physicians, acknowledging that human clinicians provide judgment, empathy, and trust that AI cannot replicate.
Experts such as MIT’s David Sontag praised the methodology behind the project but urged caution. They noted that doctors in the study were not allowed to use typical diagnostic resources and said the AI’s cost-saving potential needs to be proven through real-world clinical trials. Eric Topol of the Scripps Research Institute described the results as impressive, particularly given the complexity of the cases involved.
Microsoft has not yet confirmed plans to commercialize the technology, but potential uses include integrating it into Bing for consumer health queries or developing professional tools to assist medical practitioners.