The People's Pharmacy · Health & Longevity · November 2026
A host noted that AI answers can be both marvelous and wrong. Greene, a pediatrician who runs an AI health site, says the error rate on health questions is still significant, and gives his rule for using it.
…source now. Even the ones who go to search engines like they always did, it's that little AI summary at the top they're reading rather than going to the original source. And the answers are often marvelous, but they're often wrong, too. Even I get fooled by them from time to time. I was skeptical.
They could be both marvelous and wrong at the same time.
And confident and marvelous and wrong. And there are different studies looking at what the error rate is, but it's still significant. And on health, you don't want to make those mistakes. So I tell people to be skeptical or fact-check everything somewhere else. But I tell people that AI is a great place to start a search, but it's not a great place to end one, to start a journey into learning something, but don't take it at face value. Always fact-check it elsewhere with somebody else. And then today we were at a meeting of some of the founders of the Society for Participatory Medicine together. And this marvelous idea, think of AI. Use AI as a sparring partner, somebody to test each other back and forth, not as a crutch.
I'm wondering what worries you most about patients using AI.
The errors do concern me that people will trust false information. But I think more deeply, I'm concerned about passivity. That people could use it as a crutch and use it to get a quick answer without thinking about it or without taking actions that get their bodies aligned with the way that…
I often wonder if this Massive amount of information that keeps coming in can sort of enhance this idea of the helicopter pilot that, you know, parents are hovering over their children, worrying about their children, overly concerned because everything they do, they're questioning on AI to say, is…