STAT+: Clinical chatbots are taking medicine by storm. Should doctors trust them?

Clinical large language models developed by companies like OpenEvidence, Doximity, and UpToDate are widely used by U.S. doctors, but a June 2024 study in Nature Medicine found these models performed worse than generalist AI models on clinical questions. The findings sparked significant debate within the health AI community, with some interpreting the results as a 'clear victory for general frontier models.'