Mustafa Suleyman, CEO of Microsoft’s AI arm, says the tech giant has taken “a genuine step toward medical superintelligence,” Wired reported June 30.
In a recent test, Microsoft’s new AI system — called the MAI Diagnostic Orchestrator — correctly diagnosed patient cases 80% of the time. The experiment tested whether the tool could correctly diagnose a patient with an ailment, mimicking work typically done by a human physician.
According to the company, the results from the tool are four times better than the human physicians in the study, who reached the correct answer 20% of the time.
The tool doesn’t rely on just one system. Instead, it brings together several AI models — including OpenAI’s GPT, Google’s Gemini and others — that work together like a team of medical experts debating a case. Microsoft tested the tool using more than 300 clinical case studies from the New England Journal of Medicine.
Along with diagnosing more accurately, the AI system also saved money by picking less expensive diagnostic tests. Microsoft said it reduced costs by about 20% in simulations.
Microsoft has not decided if the tool will be used in hospitals or made available to the public, but one possibility is adding it to Bing’s search engine to help people understand symptoms. Before that, experts say it needs to be tested in clinical trials to evaluate how it performs in real-life care.