Google DeepMind unveiled its AI co-clinician research initiative to address the global healthcare worker shortage predicted to reach 10 million by 2030. This approach envisions "triadic care" where AI agents assist patients under physician supervision, extending clinical expertise while maintaining human judgment. The research builds on DeepMind's medical AI journey from MedPaLM knowledge tests to AMIE's simulated consultations.

Clinical evaluations showed AI co-clinician outperforming leading evidence synthesis tools in head-to-head blind tests. In analyzing 98 primary care queries, the system recorded zero critical errors in 97 cases. On the challenging OpenFDA RxQA benchmark for medication knowledge, the AI demonstrated superior performance when handling open-ended questions as clinicians would encounter in practice.

Beyond text-based interactions, DeepMind is exploring multimodal capabilities using live audio and video. Simulation studies with 20 clinical scenarios demonstrated the AI's ability to guide physical examinations, such as correcting inhaler technique. While still outperformed by physicians in identifying red flags, the AI matched or exceeded primary care performance in 68 of 140 assessed consultation skills, showing concrete promise as a clinical support tool.