2026-10-12→13 (Pichai/Hassabis posts, via the ai-social window): Google's AMIE published in
The Lancet (BIDMC Medicine partnership) — the first prospective study of patient-facing
conversational diagnostics in a real clinic: clinicians reported the summaries helped prepare
for visits in 75% of cases, influenced their approach in more than half, and AMIE's
differential diagnoses matched the doctors' final diagnoses 90% of the time. The
institutional twin of the Nolla prescribing thread — opposite evidence cultures (prospective
peer review vs consumer access). Watch: any product surface, independent clinical read-outs,
convergence with the prescribing-oversight arc.
x.com/demishassabis
2026-10-05 opened (Verge per Bloomberg, 4:14 PM ET): Nolla Health
launched in Utah — scan your face, the AI grades acne severity and
"autonomously writes a prescription"; two physicians approve each
AI-generated prescription for the first 100 patients, with
physician oversight that "gradually loosens." Adjacent to OpenAI's
ChatGPT-for-Clinicians vertical (Sep 10) — access first, autonomy
next. Watch: expansion beyond Utah/acne, the approval step thinning,
any state medical- or prescribing-board response, patient outcomes.
Background
The thread: AI systems writing prescriptions directly for consumers
— with the pilot design (human approval for the first cohort,
oversight loosening from there), the decay curve on the human check
is itself the story.