Only the Top 1 Percent: A Tech CEO Says a Chatbot Beats the Other 99%
For the best insured patient, the model competes with the top one percent. For millions of others, it competes with no one.
A tech CEO says he would only trust the top 1 percent of doctors, and that a chatbot beats the rest.
He left out the bigger number: the millions of patients for whom the chatbot competes with no doctor at all.
The studies, and the access data, back the direction.
Dario Amodei runs Anthropic, one of the largest artificial intelligence companies in the world. On Stripe’s “A Cheeky Pint” podcast, in conversation with John Collison, he repeated something Nobel Prize winning biologists had told him. They said they would only see the top 1 percent of doctors. For the other 99 percent, a large language model gives better advice. Amodei agreed. He said it really is true.1
That is a hard sentence to read if you wear a white coat. I have worn one for fifty years.
So let me be precise about what the claim is and what it is not.
It is not that a chatbot can deliver your baby, set your fracture, or run a code at three in the morning.
It is a claim about the cognitive work that fills most of a clinic day. Reading a history. Putting many facts together. Matching a pattern you have seen a thousand times. Staying consistent when you are tired, rushed, and behind. Amodei’s reasoning was simple.
Medicine is full of pattern matching and synthesis. Doctors are overworked and human.
Machines do not get tired, and they do not skip a step because the waiting room is full.
The reason this is worth your attention is not that a tech executive said something provocative. It is who he was quoting. The people voting for the machine here are not patients who do not know better. They are Nobel laureates, the people with the most knowledge and the most access to elite care. When the best informed patients start routing around the average doctor, that is a signal.
There is a second comparison hiding inside that sentence, and it matters more than the first.
The Nobel laureate is weighing the model against the best doctor money can buy.
Millions of Americans are weighing it against no one.
No appointment for months.
No clinician within an hour’s drive.
No insurance card in the wallet.
For them the real question is not whether a chatbot beats the average doctor. It is whether it beats nothing, because nothing is what they have.
Both claims are testable.
So I tested them against the data.
The data is more uncomfortable than the quote. Paid subscribers read on.
