Google DeepMind's AI Co-Clinician Outperforms GPT-5.4 in Doctor Tests But Still Trails Experienced Physicians

A blind evaluation of 98 primary-care queries has positioned Google DeepMind's experimental AI co-clinician ahead of OpenAI's GPT-5.4-thinking-with-search by a margin of 63 to 30 in physician preference, yet both systems still fell short of experienced human doctors on critical safety tasks such as

📖 Read the Full Brief