Microsoft says its AI is more accurate than doctors at diagnosis: what a peer-reviewed Science study showed, and what it did not
On October 22, 2026, Microsoft published a peer-reviewed study of its Diagnostic Orchestrator in the journal Science: on archival cases, it diagnosed more accurately than doctors. Based on The Verge and the Science paper itself, I break down which numbers the peer review actually backs, what the «4 times more accurate» claim is built on, and why testing on past cases is not the same as working in a real clinic.
Read more: Microsoft says its AI is more accurate than doctors at diagnosis: what a peer-reviewed Science study showed, and what it did not



