Anyone else notice this with Health AI?

The moment a new tool works, the reference point moves. Five minutes ago the problem was intractable. A new solution (quite impressive) enters the scene. And suddenly all the focus is on the gap between 70% and 100%, and the tool that closed the first 70 starts to feel like a disappointment.

I see it constantly. Ambient scribes get scrutinized for the one line a clinician has to verify, while everyone forgets that the pre-scribe note was copied forward at 11 PM and unreadable, or late and contributing to lagging AR. A text-message engagement tool reaches 40% of patients who were getting nothing, and the question from the audience is about the 5% who don’t have phones.

Holding AI to a higher bar than we hold each other is, on balance, a good instinct. But we need just a little calibration. The best comparison isn’t tool today versus imagined perfection. It’s tool today versus the status quo we were drowning in five minutes ago.