In real-world test, an AI model did better than ER doctors at diagnosing patients

NPR reports on a real-world study in which an AI model outperformed emergency room doctors in diagnostic accuracy, with the model scoring over 11% higher than human physicians on test cases. The article is authored by Will Stone for a major national outlet with strong editorial standards, lending credibility to the reporting, though the provided metadata does not include specific details about the research methodology, sample size, or the particular AI model tested. Coverage from Gizmodo and Science News corroborates the headline finding while adding critical context: the Gizmodo piece emphasizes caution about the results, and Science News notes that AI still requires real-world validation and human oversight before it can guide patient care—nuances that suggest the NPR story should be read alongside these more measured takes. Watch for peer-reviewed publication of the underlying study, regulatory commentary from medical bodies, and follow-up reporting on whether these results replicate in broader clinical settings or if the test cases had selection bias favoring AI performance.