Skip to content
See the World Through ScienceA project of ALLATRA
Source: Peer-reviewedThe Lancet1 source

Google's Medical Chatbot Interviewed Real Patients Before Their Doctor Visits

Science

Republish this story

Our work is licensed under Creative Commons BY-NC 4.0. You may republish this piece for free — with credit to ALLATRA Media and a link to the original, unedited beyond length trims, and not for commercial use.

Read the full license

Street view of the glass and brick buildings of Beth Israel Deaconess Medical Center in Boston, with cars and a walkway at the main entrance.
The Boston hospital where the study ran. Patients talked with the chatbot before an urgent appointment at one of its primary care practices."Beth Israel Deaconess Medical Center (BIDMC) (54953356537)" by ajay_suresh, via wikimedia, CC-BY-4.0

Patients at a Boston primary care practice talked with a Google medical chatbot before real urgent appointments, and not one of those conversations had to be stopped on safety grounds, according to a study published Oct. 8, 2026, in The Lancet. Alphabet, Google's parent company, funded the work, and most of the authors work for Google or Google DeepMind.

The authors say they set out to measure the safety and quality of an AI system's conversations with patients in genuine urgent appointments rather than in simulation. The paper describes a single-arm design, which means there was no comparison group: it can report how the supervised sessions went, but not whether the system matched or outperformed a clinician.

The study ran at one Beth Israel Deaconess Medical Center practice from April to November 2025. English-speaking adults used the system, called Articulate Medical Intelligence Explorer, or AMIE, up to five days before an urgent appointment for a single complaint. Physician safety supervisors monitored every interaction and were trained to step in on criteria set in advance. AMIE's transcript and summary went to the patient's own doctor before the visit. Of 114 patients enrolled, 98 finished both the chatbot session and the appointment.

The researchers report that no conversation met the prespecified criteria for a safety stop, an observation in a small supervised group rather than evidence about the system at scale. Supervisors noted one hallucination and added clinical information in five interactions. Clinical evaluators rated the conversations favorably on 17 criteria, in 87% to 100% of cases depending on the criterion, and patients rated them favorably in 48% to 96% of cases. Both sets of ratings are the study's own assessment of a system built by its funder.

In 44 of the cases, the patient's doctor had read AMIE's transcript before the visit, and in 33 of those the doctor reported afterward that it helped with preparation. The authors call the result initial feasibility and say further research is needed. It is registered with ClinicalTrials.gov as NCT06911398.

Sources

Medical Disclaimer: This content is provided for general informational and educational purposes only. It is not medical advice and should not be used as a substitute for professional medical advice, diagnosis, or treatment. Always consult a qualified healthcare professional regarding any medical condition, concern, symptom, or treatment decision.

Spot an error?

Spot an error?

Report an error

Spotted a mistake on this page? Tell us what's wrong and our editors will take a look.

What kind of problem?

Only if you'd like us to be able to follow up. We won't use it for anything else.

We correct mistakes openly. Select any text to flag it. Fixes are logged under our Corrections Policy.

Report an error

Reporting on

Google's Medical Chatbot Interviewed Real Patients Before Their Doctor Visits

What kind of problem?

Only if you'd like us to be able to follow up. We won't use it for anything else.

We read every report. Corrections are logged publicly.