analiza aiqbench
This study demonstrates that orchestration layers like Microsoft's MAI-DxO can significantly enhance diagnostic accuracy and reduce costs when paired with advanced foundation models such as GPT-5. However, the performance gains are driven primarily by the underlying LLM capabilities rather than the specific coordination logic of the system itself. This highlights a critical trend where infrastructural improvements in model efficiency are becoming the dominant factor in clinical AI applications.
Microsoft’s Mustafa Suleyman says its AI outperforms doctors on hard cases as he chases ‘medical superintelligence’
Microsoft’s new AI diagnostic system beat generalist doctors on hard cases, according to a study in Science. But most of the improvement seems to come from the AI models underneath, not from Microsoft’s system itself.
Microsoft’s AI diagnostic system correctly diagnosed complex medical cases in a simulation more than twice as often as generalist physicians, according to a study published Thursday in Science. Mustafa Suleyman, Microsoft AI’s CEO, called the result a step toward “medical superintelligence.” The system, dubbed MAI-DxO (DxO for short), coordinates several AI agents that act like a panel of doctors. They question a patient, order tests, and debate a diagnosis. In a simulation using complex cases from the New England Journal of Medicine, it was correct 83% of the time. Primary care doctors and general internists, working individually with reference materials, were correct 36% of the time. One note: cases this complex are usually handled…
Extras afișat conform termenilor RSS ai publicației · 114 cuvinte · textul integral și figurile sunt pe site-ul publicației
Subiecte conexe
Microsoft says its AI tool is better than doctors at diagnosing conditions
The Verge · 22 oct. 2026
Superintelligent Retrieval Agent: The Next Frontier of Agentic Retrieval
arXiv · 8 iun. 2026
AI Companies Could Become More Powerful Than Their Host Nations
MIRI · 17 iun. 2026
Toward Training Superintelligent Software Agents through Self-Play SWE-RL
arXiv · 3 iun. 2026