The Information Machine
Following·Day 5·in the edition since 13 Oct 2026·2 sources·updated 13 Oct 2026

Google AMIE chatbot's Lancet clinical trial

The gist

Google AMIE Chatbot Hits 90% Diagnostic Accuracy in Lancet Clinical Trial

A Lancet-published trial with real patients offers harder evidence for clinical AI performance than benchmark testing alone. The absence of a comparison group and the supervised setting leave open how AMIE would perform in unsupervised clinical conditions.

How it developed
First in the edition · 13 October 2026
12 October 2026

The Lancet published results on October 12 from a supervised trial of Google's AMIE, a chatbot that collected symptom histories from 98 urgent care patients in Boston before appointments; the physician's final diagnosis appeared in AMIE's differential 90% of the time.

A physician monitored every conversation: item 11582 reports supervisors caught one hallucination and added clinical information in five chats, while item 11643 states they never needed to intervene. The trial had no comparison group, and item 11582 notes the results do not establish how AMIE would perform without human supervision.

Sources
The daily email

Want this in your inbox?

I send one email each morning with the stories that moved. If you would rather just read here, that works too.

Subscribe free