Pew Research asked more than three thousand Americans in June 2026. Thirty-four percent use AI chatbots for a health reason. About a quarter use them to diagnose symptoms.
A separate poll from KFF found that among people who used AI for health, one in five said a major reason was that they couldn't afford to see a provider.
What's the pattern?
Now here's the number nobody puts next to it.
In January 2013, Pew reported that 35 percent of Americans had gone online to figure out a medical condition. They called them "online diagnosers." Half of them followed up with a doctor. And the doctor confirmed the guess more often than not.
Thirty-five percent in 2013. Thirty-four percent in 2026.
The share of people who look before they call hasn't moved in thirteen years.
What moved is the quality of the answer. And the reason they're asking. In 2013 it was curiosity. In 2026, for one in five, it's cost.
Before the search bar there was the medical encyclopedia on the shelf. Before that, the pharmacist. Patients have always triaged themselves. Medicine has always panicked about it. And then medicine absorbed it.
So where's the real risk?
Not in the chatbot being wrong. Doctors have been correcting wrong self-diagnoses since the encyclopedia.
The risk is the patient who gets a confident answer and never makes the call. In 2013, half followed up. The number to watch now is the half that doesn't, especially the ones who asked a machine because the visit costs too much.
That's not a technology failure. That's an access failure wearing a technology costume.
What should leaders do now?
If you run a health system, three things.
Assume the patient arrives with a draft diagnosis. Design the visit as a review, not a blank page. It's faster, and it's honest.
Meet the one in five. If cost is why they asked a bot, then a chatbot inside your own system that ends with a real human triage line is the product.
And train for the "no." The most valuable skill in an AI-first intake is the confident correction, delivered without contempt for the person who asked a machine first.
The doctor was never the first opinion. The doctor was always the one that counted.
That's still true. If we build for it.
When did you last look something up before you made the call?
Related: Will we live to 120, and who pays for it?
More like this on LinkedIn. Follow @samradofficial
