How do patients find a doctor with AI now?

Ten years ago, a patient with a torn meniscus opened Google, typed "knee surgeon near me," and worked down a page of ten links until something looked right. That patient is disappearing. The one replacing them opens ChatGPT, asks "who's the best knee surgeon in Denver," reads the three names it gives back, and calls one. No page of links. No scrolling. No comparison shopping. Three names, and a decision.

This is not a prediction about the future. The number of patients using AI to find a doctor roughly doubled in nine months, and for a growing share of them the AI answer is not the start of their research, it is the end of it. Whoever the assistant names gets considered. Everyone else is not sitting in a lower spot on the list. They are not on the list at all, because there is no list anymore.

For a physician, that should be alarming and clarifying at once. Alarming, because you can be an excellent surgeon in your city and simply never come up. Clarifying, because "being found" now has a specific, checkable meaning, and it is not the one most practices are still chasing.

The old game and the new game are being played at the same time

Google did not vanish, and ranking still matters. But a second game opened on top of the first, and it does not work the same way. In the old game you competed for a position on a page of results. In the new game you compete to be one of a few names inside a single answer, and the thing that decides that answer is not the thing that won you a Google ranking.

That is why good practices get blindsided. They spent years and real money winning the old game and assumed it carried over. It does not. A strong Google position and a polished website are answering a question fewer and fewer patients are asking.

Two different questions, two different answers

There is a split here that decides which of your content actually helps you. When a patient wants to understand a condition, "what does a torn meniscus feel like," an AI pulls from general health references and explains. When a patient wants a person, "who should I see for a torn meniscus in Denver," it does something completely different: it assembles a short list of real physicians it can confirm are real, nearby, and known for that specific thing.

Most practices pour their effort into the first kind of content, the condition explainers, and then wonder why it never turns into patients. It never will, because it answers the wrong question. The person reading about what a torn meniscus feels like is not choosing a surgeon yet. The person asking who to see is, and that answer gets built from somewhere else entirely.

What actually decides whether you are named

An AI recommends a physician it is confident about, and confidence comes from agreement. It wants the same facts about you, said the same way, in enough trustworthy places that it feels safe recommending you to a frightened stranger. Your website is one voice in that. The louder voices are your listings across the directories, the reviews that say what you actually treat, and whether your name, specialty, and location line up everywhere someone looks. When they agree, you are easy to recommend. When they fight each other, the safe move for the assistant is to name someone whose story is clean. If you want the full picture of what tips that decision, it is worth understanding why an AI recommends some doctors and skips others.

See where you stand in five minutes

Open ChatGPT right now and ask it, the way a patient would, who the best physician in your specialty is in your city, and who does your signature procedure nearby. Ask two or three times, because the answer moves. Write down whether your name appears at all, and whose does. That five minutes will tell you exactly where you stand in the game that is quietly deciding who gets the call, and it may be the most useful five minutes you spend on your practice this month.