Model
119d agoDeepMind Co-Clinician Outperforms GPT-4 in Medical Tests
Google DeepMind developed an AI co-clinician that beats GPT-4 in blind tests conducted by doctors. Despite this edge, the system still trails experienced human physicians in diagnostic accuracy. The study also highlights why ChatGPT's voice mode fails in high-stakes medical consultations.
The Signal
Practitioners should view these tools as assistants rather than autonomous diagnostic replacements.