Microsoft has released early research on an experimental system called the Microsoft AI Diagnostic Orchestrator, or MAI-DxO, that it says can diagnose complex medical cases far more accurately and cheaply than experienced physicians. The company reported that the tool correctly solved roughly 85% of 304 challenging case reports drawn from the New England Journal of Medicine, compared with a 20% success rate for a panel of 21 doctors working under the same constraints.
MAI-DxO works by orchestrating up to five specialised AI ‘agents’ that debate possible illnesses, order diagnostic tests and converge on a final answer. The orchestrator taps several frontier language models—including OpenAI’s GPT, Google’s Gemini, Anthropic’s Claude, Meta’s Llama and xAI’s Grok—to simulate a multidisciplinary clinical team. Microsoft says the approach not only boosted accuracy but lowered the cost of recommended tests by about one-fifth relative to doctors and to single-model baselines.
Mustafa Suleyman, chief executive of the company’s recently formed Microsoft AI unit, called the study a “step toward medical super-intelligence”, while emphasising that the technology is designed to assist, not replace, clinicians. The findings are being submitted for peer review, and the prototype has not undergone regulatory scrutiny or trials in live clinical settings.
The announcement underscores an intensifying race among big tech groups to commercialise generative-AI tools for health care. Microsoft recruited several senior researchers from Google to build the system and says it will spend the next few years validating the technology with hospitals and regulators before deciding on commercial deployment.
Dubbed "Microsoft AI Diagnostic Orchestrator", the new AI-powered tool was trained on 304 studies that described some of the most complex cases solved by doctors. #Microsoft #MedicalAI #AI
The technology giant Microsoft is claiming it has developed an artificial intelligence system that can solve complex diagnostic challenges better than most doctors
ScienceSoft Raises the Bar for AI Voice Scheduling in Healthcare #MachineLearning #AdversarialAI #SmartAI #AI #EnterpriseAI #GenerativeAI #GenAI #MLAlgorithms #DeepLearning #AIResearch #AIRevolution
