Beth Israel Tested Google Medical Chatbot
An eight-month clinical trial evaluated how AI-driven tools perform when assisting primary care physicians with patient cases.
Updated on Oct. 8, 2026 in Stroke

Live Poll
Would you trust medical information provided by an A.I. chatbot during a doctor's visit?
Beth Israel Deaconess Medical Center completed an eight-month clinical trial in 2025-2026 investigating the use of a Google medical chatbot named AMIE. The study, which included nearly 100 participants, assessed how Gemini-powered technology integrates into primary care workflows.
Why it matters
The trial aimed to determine if specialized artificial intelligence could assist doctors in managing patient health without providing harmful or inaccurate information. Understanding these performance metrics is essential for integrating generative AI into standardized clinical practices.
Nearly 100 patients participated in the study where the chatbot hallucinated once and physicians added context five times. Primary care doctors reported the technology was helpful in 75 percent of the observed patient cases.
The players
Beth Israel Deaconess Medical Center
This Boston-based teaching hospital is a major affiliate of Harvard Medical School and a leader in medical research.
This multinational technology company develops advanced artificial intelligence models, including the Gemini system.
The Lancet
This is a weekly peer-reviewed medical journal that publishes original research, review articles, and editorials.
The details
During the trial, patients discussed their ailments with the AMIE chatbot before meeting with their doctor, who then reviewed a summary generated by the system. Physicians noted that this technology influenced their handling of more than half of all patient cases, though they never found it necessary to halt a conversation.
Timeline
The clinical trial took place over an eight-month period between 2025 and 2026.
Study results were published in The Lancet on October 8, 2026.
The Big Picture
This study marks a significant milestone in the AMIE research program by validating the integration of LLMs in real-world clinical environments. It moves medical AI beyond theoretical testing, establishing a baseline for how automated summaries influence physician decision-making.
While these tools are currently limited to research settings, they may eventually lead to faster initial assessments and more concise doctor visits. Patients should note that doctors remain in control of these systems, as physicians manually reviewed all AI-generated summaries in this trial.
The takeaway
Artificial intelligence tools are increasingly capable of assisting physicians with patient triage and summary generation. Future healthcare interactions may become more efficient as doctors adopt these systems to support their diagnostic workflows.
Further reading
Learn more about the latest developments in medical diagnostics and care at the United States Stroke portal.
Live Poll
Would you trust medical information provided by an A.I. chatbot during a doctor's visit?








