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

A landmark study published in the journal *Science* has found that an AI reasoning model developed by OpenAI successfully outperformed emergency room doctors in diagnosing complex medical cases. The researchers, based at Harvard Medical School and Beth Israel Deaconess Medical Center, ran a series of 'real-world' tests where the AI was tasked with identifying ailments from patient records. In one notable instance, the AI correctly identified a case of lupus-related heart inflammation after human doctors had mistakenly suspected a common pulmonary embolism.

The findings suggest that AI 'co-pilots' could soon become a standard tool in triage and diagnostic decision-making. While doctors often rely on intuition and recent experiences, the AI was able to scan vast amounts of historical data and identify stable statistical patterns that humans missed. However, medical experts emphasize that the technology is intended to augment rather than replace human physicians, especially as the 'black-box' nature of AI models makes it difficult for them to explain the medical logic behind their conclusions to patients.