Model5h ago
DeepMind Co-Clinician Outperforms GPT-5.4 In Medical Tests
Blind doctor tests show Google DeepMind's AI co-clinician outperforms GPT-5.4 in clinical simulations. Despite the win, the system still trails experienced physicians in diagnostic accuracy. The study also highlights critical failures in ChatGPT's voice mode for medical tasks. These gaps suggest AI remains a supportive tool rather than a primary provider.