Microsoft introduced its "AI Diagnostic Orchestrator" (MAI-DxO), claiming it achieves an 85.5% diagnosis accuracy on complex medical cases, surpassing experienced physicians, who scored only 20%. Developed by Mustafa Suleyman’s AI health unit, the tool, trained on 304 studies, employs a "chain of debate" technique involving five specialized AI agents simulating expert roles. Microsoft’s CEO emphasizes this innovation as a stride toward "medical superintelligence," though further validation with healthcare organizations is necessary before broad deployment.
Microsoft introduced its "AI Diagnostic Orchestrator" (MAI-DxO), claiming it achieves an 85.5% diagnosis accuracy on complex medical cases, surpassing experienced physicians, who scored only 20%. Developed by Mustafa Suleyman’s AI health unit, the tool, trained on 304 studies, employs a "chain of debate" technique involving five specialized AI agents simulating expert roles. Microsoft’s CEO emphasizes this innovation as a stride toward "medical superintelligence," though further validation with healthcare organizations is necessary before broad deployment.