Google's AMIE matches primary care physicians in real-time video clinical consultations

Google Research has extended its AMIE (Articulate Medical Intelligence Explorer) diagnostic AI system to conduct real-time audio-visual consultations, and a new study finds clinical evaluators rated its performance favorably against practicing primary care physicians.

What's new

Google announced on its official blog on August 11, 2026 that AMIE now "guides virtual physical exams, and reasons diagnostically in real time" over live audio and video, rather than text chat alone. The system was tested in "a randomized study using simulated consultations with patient actors and a group of primary care physicians," and Google reports: "clinical evaluators assessed AMIE favorably across core clinical competencies, including history-taking thoroughness, diagnostic accuracy, management appropriateness and communication quality." The companion Google Research technical post, describing the same study across 100 clinical scenarios and 300 live consultations rated by 30 board-certified physicians, adds that AMIE was rated especially strong at eliciting physical signs and guiding patients through virtual examinations, and that "patient actors also preferred the video experience over text chat."

Google is explicit about the study's limits: AMIE "remains a research system and more research is needed before responsible real-world clinical deployment." The evaluation used professional patient actors in simulated settings rather than real patients presenting with genuine health conditions.

Context

AMIE has progressed over several years from text-only diagnostic dialogue to a system that reasons over medical images and now live video and audio, architected around separate agents for patient-facing conversation, clinical reasoning, and non-verbal perception. Google is running a separate real-world study with Beth Israel Deaconess Medical Center to assess AMIE prospectively outside simulated conditions, underscoring the gap the company itself draws between actor-based evaluation and genuine clinical deployment.

Why it matters

Much of the AI-in-medicine conversation has centered on text or image-based diagnostic accuracy, but a large share of real telemedicine happens over video, where non-verbal cues, guided self-examination, and rapport matter as much as diagnostic reasoning itself. AMIE's video version outperforming the text-only version specifically on physical-examination guidance suggests the video modality captures information a text-based system structurally cannot — a meaningful data point for telehealth providers evaluating whether video-based AI triage could extend, rather than just replicate, chat-based tools. By foregrounding the actor-versus-real-patient gap itself, Google is framing this as a research milestone rather than a readiness claim for clinical deployment.