How do patients find doctors through ChatGPT and other AI assistants?

How do patients find doctors through ChatGPT? They describe a need and a city, and the assistant runs several web searches, reads directories and practice websites, then names a handful of providers with reasons. In our US panel, ChatGPT named specific providers in all 40 answers, but the names changed from one run to the next.

Patient asking an AI assistant on her phone to recommend a local doctor

How do patients use ChatGPT to find a doctor?

A patient asks something like "who is a good dermatologist in Austin for adult acne?". The assistant searches the web in the background, often targeting directories and board-certification sources, reads the results and answers with several named providers and links. Asking again often returns a different list.

A growing share of patients start with an AI assistant rather than a search box. KFF's Health Misinformation Tracking Poll (August 2024) found that about one in six US adults (17%) said they use AI chatbots at least once a month to find health information and advice, rising to 25% of adults under 30. The same poll found most adults (56%) were not confident that health information from AI chatbots is accurate. So patients use these tools, and they still check.

That poll asked about health information in general, not specifically about choosing a doctor. To see what happens when the question is "who should I see?", we ran our own test.

What did we test, and how?

On 19 November 2026 we asked five questions phrased the way patients ask them, each with a specialty, a need and a US city: a dermatologist in Austin for adult acne, a plastic surgeon in Miami for a breast lift consultation, a dentist in Denver for dental implants, a chiropractor in Phoenix for low back pain, and an orthopedic surgeon in Chicago for a torn meniscus. Each question went to ChatGPT, Perplexity, Gemini and Google AI Mode 8 times, through DataForSEO with web search on and a US location where supported. That produced 40 answers each from ChatGPT, Perplexity and Gemini and 35 from AI Mode (a sixth question and some AI Mode runs failed when our data account ran out of credit).

API answers approximate the consumer apps but are not identical, five cities is a small sample, and provider names were extracted automatically, so the figures are approximate. The complete method is in our AI search visibility benchmark.

What does ChatGPT do when a patient asks for a doctor?

It searches before it answers. ChatGPT exposed its background searches in all 40 of its answers, an average of about five searches per question. Those searches show what it is looking for:

  • Directories by name. In 38 of 40 answers at least one background search was restricted to, or named, a directory or professional society, such as healthgrades.com, zocdoc.com or the American Academy of Dermatology's dermatologist finder.
  • Credentials. In 20 of 40 answers a background search included "board certified".
  • Reviews. In 18 of 40 answers a background search included the word "review".

This process is often called query fan-out: one patient question becomes several narrower searches. Gemini showed its background searches too, averaging about two per question. The practical point is that the assistant goes looking for independent confirmation of who you are and what you are qualified to do, and it looks in specific places.

Do AI assistants actually name specific doctors?

Usually, yes. Answers that named at least one specific provider or practice:

EngineAnswers naming a specific provider95% interval
ChatGPT40 of 40 (100%)91 to 100%
Perplexity40 of 40 (100%)91 to 100%
Gemini39 of 40 (98%)87 to 100%
Google AI Mode21 of 35 (60%)44 to 74%

ChatGPT, Perplexity and Gemini nearly always produced a short list of named practices with reasons, often grouped by what each is known for. Google AI Mode was more reserved: in many answers it pointed to map results and reviews rather than naming providers in the text. Most answers also included a caution to confirm insurance, credentials and availability directly with the practice.

Which sources do the assistants cite?

Two kinds of source dominate: practice websites and doctor directories. Practice websites were cited in 100% of Perplexity and Gemini answers, 74% of AI Mode answers and 53% of ChatGPT answers. A directory or review platform was cited in 80% of Perplexity answers, 65% of ChatGPT answers, 43% of Gemini answers and 23% of AI Mode answers.

Each engine had its favorites. ChatGPT cited Healthgrades in 16 of 40 answers; Perplexity cited WebMD doctor profiles in 24 of 40; Gemini cited Reddit in 11 of 40 and RealSelf in 8; Google AI Mode cited Google Maps or Business listings in 32 of 35. We break this down platform by platform, with a checklist, in which directories AI assistants use for doctors.

Why does the answer change every time a patient asks?

Because the assistant rebuilds the answer each time: it reruns the searches, gets slightly different results, and samples its wording. We measured how much the named providers overlap between two runs of the same question, on a scale from 0 (no names in common) to 1 (identical lists).

EngineAverage overlap between two runs, per questionWhat it means
ChatGPT0.01 to 0.09Two runs rarely share more than one or two names
Gemini0.05 to 0.26Some names recur, most do not
Perplexity0.18 to 0.54The most stable: a core group of practices recurs
Google AI Mode0.01 to 0.13Short lists that shift between runs

For a practice, this changes what "being recommended" means. A single screenshot proves very little; one patient might see your name and the next might not. The useful measure is how often you appear across many runs, which is why we ask each question 8 times per engine and report a rate with an interval. You can run a lighter version yourself with our AI visibility checker.

What do patients actually ask?

Patient questions to assistants tend to be longer and more specific than search queries. They combine a condition or procedure, a location, and often a preference: "a dentist in Denver who does implants and takes my insurance", "a dermatologist who is good with adult hormonal acne", "a surgeon who does a lot of revision cases". The more specific the question, the more the assistant needs pages that state those specifics plainly. A practice site that says only "comprehensive care for the whole family" gives the assistant nothing to match.

What should a practice do about it?

  1. Say who you treat, where, in the first paragraph of your home page and each service page. Assistants extract plain statements far more easily than slogans.
  2. Give each service or condition its own page, written around the questions patients ask, and reviewed by the physician.
  3. Make credentials checkable. Board certification, training and hospital affiliations on the physician bio, matching what directories and specialty societies show.
  4. Fix the directories assistants search. Google Business Profile, Healthgrades, WebMD, Zocdoc if you use it, your specialty society finder. Same name, address and specialty everywhere.
  5. Build genuine reviews through a steady routine that asks every patient the same way, without incentives or filtering, as Google's policy and the FTC's review rule require. See Google reviews for doctors.
  6. Measure across repeated runs, not with one-off screenshots.

None of this guarantees a mention. AI answers are not for sale, and they vary by design. These steps make your practice easier to find, verify and quote, which is what the assistants are visibly looking for. Our guide to getting your practice recommended by ChatGPT goes deeper on the website side.

Who ran this test?

Only One Medical is a marketing agency for physicians based in Israel, working remotely with US practices on SEO, Generative Engine Optimization (GEO) and websites. Our client work so far is in Israel. We ran and paid for this panel ourselves; no platform sponsored it. How we count our own numbers is on our numbers page, and our service for practices is described on GEO for medical practices.

Sources

Frequently asked questions

Do patients really use ChatGPT to find doctors?

Some do. KFF's August 2024 poll found 17% of US adults use AI chatbots at least monthly for health information and advice, 25% of those under 30. The poll did not ask specifically about choosing a doctor, so treat it as a lower-bound signal of habit, not a measure of doctor searches.

Does ChatGPT recommend specific doctors?

In our November 2026 US panel it named at least one specific provider in all 40 answers. It also usually advised patients to confirm credentials, insurance and availability with the practice.

Where does ChatGPT get its information about doctors?

From live web searches. In our panel its background searches targeted directories such as Healthgrades and Zocdoc in 38 of 40 answers, and it cited practice websites in 53% of answers and Healthgrades in 40%.

Why did ChatGPT recommend a different doctor when I asked again?

Each answer is rebuilt from fresh searches. Between two runs of the same question, the provider lists in our panel overlapped very little for ChatGPT (an average overlap of 0.01 to 0.09 on a 0 to 1 scale).

Can a practice pay to be recommended by ChatGPT?

No. There is no paid placement inside ChatGPT's answers. A practice can only make itself easier to find and verify: clear website pages, accurate directory profiles, genuine reviews.

How can I check whether AI assistants mention my practice?

Ask the same patient-style question several times in each assistant and count how often you appear. Our free AI visibility checker generates the questions and a tally sheet.

Want to know how AI assistants see your practice today?

Send us your website. We will check how ChatGPT, Perplexity, Gemini and Google describe your practice and tell you, in plain English, what is holding you back. Pricing depends on scope, so we quote after the first call.

Keep reading

Citing this page. You may quote this article with attribution.Fargon, A. (2026). "How do patients find doctors through ChatGPT and other AI assistants?". Only One Medical. https://onlyonemedical.co.il/en/insights/how-patients-use-ai-to-find-doctors