The integration of artificial intelligence into clinical practice is moving beyond administrative support and entering the core of medical diagnosis. A recent study by Google Research, involving 13,917 real patients, demonstrated that the conversational agent SymptomAI can identify pathologies with higher precision than primary care physicians.

The Dynamic Approach of SymptomAI

Built on Gemini 2.0 Flash, the agent does not simply answer questions but conducts an active anamnesis interview. The system is designed to ask follow-up questions, adapt the conversation in real time, and produce a structured differential diagnosis (DDx), mimicking the logical process a physician uses during an office visit. The experiment, deployed within the Fitbit app from June 2025 to April 2026, showed an accuracy rate of 73%, compared to 60% for physicians reviewing the same transcripts.

Google's Latest (and Fastest) Gemini Model Is Now Available to Everyone ... — https://www.makeuseof.com/google-update-gemini-2-flash-general-availability/

Beyond Accuracy: Digital Bedside Manner

In addition to technical data, results suggest the chatbot scored highly in "bedside manner," interacting with patients empathetically. This is part of a broader ecosystem including Med-Gemini, focused on radiology and genomics, and Google SensorFM, which leverages wearable data to predict metabolic and cardiovascular conditions.

The Complexity Ceiling

Despite the statistical edge, a clear boundary remains: in complex clinical cases, human doctors maintain a significant lead (91% accuracy vs. 71% for AI). The ability to interpret non-verbal cues, deep patient history, and contextual intuition remains a human prerogative. Rather than replacement, the trend points toward AI as an "ideal colleague": always available, fatigue-free, and capable of mitigating the cognitive biases that can lead a physician to anchor to an initial incorrect diagnosis.

Google dévoile ses modèles Gemini 2 Pro, Flash Thinking et Flash-Lite ... — https://iphonesoft.fr/2025/02/06/google-modeles-gemini-2-pro-flash-thinking-flash-lite

Global Healthcare Outlook

The shift toward AI-driven triage could drastically reduce the burden on primary care systems worldwide. However, the transition necessitates a global consensus on liability and medical ethics to ensure that AI remains a supportive tool rather than a sole decision-maker.