Writing
Health AI is easy to build and hard to make trustworthy. These are working arguments about the difference — each one anchored in the research, aimed at the design decisions that actually carry the weight.
A blanket alert can close a liability gap overnight. What it opens — alert fatigue, and the clinical risk that follows — depends on who was in the room when it was built.
Grounded in the clinical decision support literature on alert overrides and alert fatigue, and the 2021 JAMA Internal Medicine validation of the Epic Sepsis Model.
A responsible health-AI output isn’t one a clinician just accepts—it’s one they keep double-checking, and that starts long before the screen.
Grounded in the automation-bias literature in JAMIA and human-computer interaction research on AI overreliance.
Ambient AI scribes are giving clinicians their attention back. The question is what disappears along with the paperwork — and who's left accountable when it does.
Grounded in JAMA, NEJM Catalyst, and the Auditor General of Ontario's 2026 AI audit.
Health AI keeps borrowing trust from doctors. For the people who trust doctors least, there's nothing to borrow. And that's where you find out what a product is actually worth.
Grounded in a JAMA Network Open study on consumer trust in health AI.