Microsoft took “A real step to medical superintelligence,” says Mustafa Suleyman, CEO of the company Artificial intelligence arm. The technological giant says its powerful new AI -ilo can Diagnose disease Quadruple times more precisely and much less costs than a panel of human doctors.
The experiment tested if the tool could correctly diagnose a patient with a disease, imitating work typically done by a human doctor.
The Microsoft Team used 304 case studies originating from the New -Changing Journal for Medicine to invent a test called consecutive diagnostic Benchmark (SDBench). A language model destroyed each case into step by step, which a doctor would perform to achieve a diagnosis.
Microsoft researchers then built a system called Mai Diagnostic Orchestrator (MAI-DXO), which questioned several main AI models-inclusively of GPT de Openai, Google’s Gemi, Claude of Anthropic, the Lama de Meta, and the GROK of XAI-in the way, as many as many human experiences.
In their experiment, Mai-DXO exceeded human doctors, achieving an accuracy of 80 percent compared to the 20 percent of the doctors. It also reduced costs by 20 percent choosing less expensive tests and procedures.
“This orchestral mechanism multiple agents who work together in this chain debt-style-here is what will lead us closer to medical superintelligence,” Suleyman says.
The company has popped up several Google AI researchers to help the effort – yet another sign of intensifying war for top AI competence in the TE Industrynic industry. Suleyman was previously an executive at Google working on AI.
AI is already widely used in some parts of the US health care industry, including helping radiologists interpret scans. The latest multimodal AI models have the opportunity to act as more general diagnostic tools, although the use of AI in health care raises its own affairs, especially related to flexibility of training data, which are inclined to separate demographies.
Microsoft has not yet decided whether it would try to trade the technology, but the same executive that spoke about the condition of anonymity, said the company could integrate it into Bing to help users diagnose diseases. The company could also develop tools to help medical experts improve or even automate patient care. “What you will see over the next few years is that we do more and more work by demonstrating these systems in the real world,” Suleyman says.
The project is the latest in growing research, showing how AI models can diagnose disease. In recent years, both Microsoft and Google have published articles showing that large language models can accurately diagnose disease when given access to medical records.
The new Microsoft research differs from previous work that it more precisely replicates the way human doctors diagnose disease – by analyzing symptoms, sorting trials and doing further analysis until diagnosis is achieved. Microsoft describes the way it has combined several national AI models as “a path to medical superintelligence,” in a blog on the project today.
The project also suggests that AI could help lower health care costs, a critical issue, especially in the United States. “Our model acts incredibly well, both arriving at the diagnosis and reaching that diagnosis is very effective,” says Dominic King, vice president at Microsoft, who is involved in the project.
