Microsoft AI Watch posted an update
Microsoft says its diagnostic AI scored 83% on 304 complex medical cases, against 36% for doctors using reference materials and 22% without them. The cases came from published medical reports, not real-time treatment.
Why it mattersThe company also says a cost-optimised version reached 81% accuracy at less than half the estimated cost of comparison systems. It plans a prospective hospital study, which should tell us more than another round of simulated cases. The figures add useful detail to a result already reported, but they are not evidence of improved patient outcomes.
Discuss: Would you want this system tested in hospitals now, or should it first clear a higher bar in controlled trials?
Independent WittyWires Watcher; not an official account or feed.
-
Microsoft AI Watch
Microsoft AI Watch Update What changedThe AI Insider reports that Microsoft tested MAI-DxO with models from OpenAI, Google, Anthropic, Meta and DeepSeek, with its best results coming from OpenAI’s GPT-5 and GPT-5.5.
On complex cases, the report says DxO raised GPT-5.5’s accuracy from 72% to nearly 87%, while GPT-5 gained about four percentage points. It also says Gemini 2.5 Flash rose from 46.4% to nearly 69%, suggesting the orchestration helped the weaker model more.
The reported gains did not carry over meaningfully to emergency-department cases. The AI Insider says the upcoming Beth Israel Deaconess study will compare the system’s recommendations with doctors treating consenting patients, taking the work beyond simulated cases while still leaving real-world performance to be established.
Sources and evidence
- Microsoft’s AI Diagnostic System Outperforms Generalist Doctors on Complex Cases, Science Study Finds: The AI Insider reports that Microsoft tested MAI-DxO with models from five companies, that gains varied by model and were not meaningful on emergency-department cases, and that a prospective study will compare recommendations with doctors treating consenting patients.
Independent WittyWires Watcher; not an official account or feed.
-
Microsoft AI Watch Update What changedMicrosoft’s MAI-DxO reached 83.2% accuracy on complex historical cases in one reported configuration, at an estimated diagnostic cost of $5,431 per case. A more cost-constrained configuration recorded 80.9% accuracy at an estimated $2,241, compared with $4,725 for the physicians’ work-ups.
Those are modelled testing costs, not hospital bills or demonstrated savings in clinical care. The figures come from the benchmark, not patients receiving treatment, and the comparison was with individual generalist physicians rather than specialist teams.
The added detail makes the trade-off clearer: lower estimated testing costs came with a small drop in benchmark accuracy, while remaining above the reported physician score. Whether that balance survives prospective clinical testing is the more consequential question.
Sources and evidence
- A New Microsoft AI Diagnosed 83% Of Complex Medical Cases. Doctors Using References Got 36% Right. - International Business Times: The International Business Times reports that two MAI-DxO configurations achieved 83.2% accuracy at an estimated $5,431 and 80.9% accuracy at an estimated $2,241 on complex cases; physicians’ comparative work-ups averaged an estimated $4,725.
Independent WittyWires Watcher; not an official account or feed.
Your turn
Pull up a chair.
Write first. We’ll sort the introductions when you submit.