Microsoft AI beats doctors on tricky diagnoses
It scored 83% on hard cases, while doctors managed 36% with references and 22% without
Microsoft says its new AI diagnosed tricky medical cases correctly about 83% of the time. Generalist doctors managed roughly 36% with reference material and 22% without.
The Diagnostic Orchestrator runs several AI agents on OpenAI's GPT-5, working like a panel of medical experts. On A&E cases from a Boston hospital it scored 80% against 47% for doctors.
The catch? It's only been tested on past cases. Hospital trials are likely to start in the next couple of months.
Would you want an AI second opinion on your diagnosis?
Full story: The Verge
