Short answer
Close to half of patients now use AI at some point when researching a provider, and about a third of US adults have asked a chatbot a health question. AI is one influence among several rather than the deciding one, and the bigger finding is accuracy: two thirds of AI users met wrong provider details.
Key takeaways
- 47% of patients said they have used an AI tool to research providers, up from 31% a year earlier (rater8, April 2026, n=992, vendor survey).
- Among those who searched for a provider, AI tools influenced 36%, ahead of Google at 34% and a doctor’s referral at 32%. Friends and family and the insurance portal both sat higher, at 43%.
- 66% of AI users met incorrect provider information, and 60% trusted the summary without checking it anywhere else.
- 32% of US adults have used an AI chatbot for health information, double the year before, and 81% of them acted on what it said (Rock Health, December 2025, 8,000 adults, vendor survey).
Two surveys carry most of what is known here, and both are vendor surveys rather than peer-reviewed research, so the numbers below are labelled that way every time they appear. They still agree with each other, and they point at something specific a practice can act on this month.
How many patients are actually doing this?
In rater8’s 2026 Patient Choice Report, 47% of respondents said they had used an AI tool to research healthcare providers, up 16 points from 31% the year before. The sample was 992 US adults surveyed in April 2026. Usage peaked in the 45 to 60 group at 64%, which surprises people who expect this to be a young-patient behaviour.
Rock Health’s eleventh Consumer Adoption Survey, fielded in December 2025 across 8,000 US adults, found 32% had used an AI chatbot for health information, double the 16% a year earlier. Around three quarters of those people used a general assistant such as ChatGPT rather than anything their provider or payer offered.
Is AI the thing that decides where they book?
Not on its own, and any page that says otherwise is misreading the data. Among the 714 rater8 respondents who had searched for a provider, AI tools influenced 36%, ahead of Google search at 34% and a doctor’s referral at 32%. Friends and family and the insurance portal both sat higher at 43%, and review sites at 40%.
| Influence | Share | What it means for a practice |
|---|---|---|
| Friends and family | 43% | Word of mouth still leads. Nothing here replaces it |
| Insurance website or portal | 43% | Your listing there has to be current and correct |
| Review sites | 40% | Volume and recency, from real patients only |
| AI tools | 36% | New, rising fast, and invisible in your analytics |
| Google search results | 34% | Still substantial, and no longer the whole picture |
| Doctor referral | 32% | Steady, and now below the tools patients use themselves |
One split in that data is worth noticing. Among patients who had actually switched provider, AI was the top influence at 39%. The people most likely to be shopping are the people most likely to be asking a model, which is exactly the group a growing practice wants to reach.
The finding that should worry you more than the rankings
Of the 465 rater8 respondents who had used AI to research a provider, 66% said they had encountered incorrect provider information: wrong addresses, phone numbers, insurance details or office hours. 60% said they trusted the summary without verifying it anywhere else. Wrong details do not stay online. They turn into a patient outside a locked door.
This is why the first work is almost never content. It is making your name, address, phone, hours, credentials and services agree across your own site, your Google Business Profile, the directories and the insurance listings. An assistant repeats what it can cross-check, and it cannot tell which of three phone numbers is the current one.
What people do after they get an answer
They act on it. In the Rock Health survey, 81% of AI users said they took at least one relevant action after their chatbot interaction, and 64% used health AI weekly or more. Treat these answers as a referral conversation happening without you in the room, several times a week, in your own market.
What these numbers do not tell you
Both surveys are self-reported and run by companies with an interest in the result, which is why the labels stay attached here. Neither measures what any single assistant recommends in your city, and neither separates dental from medical from allied health. The only way to know your position is to ask the questions your patients ask and write down what comes back.
Questions people ask
Do patients really pick a doctor using ChatGPT?
They use it as one input among several. In rater8’s April 2026 survey of 992 US adults, AI tools influenced 36% of people who searched for a provider, behind friends and family and insurance portals at 43%, and ahead of Google at 34%.
Which patients are most likely to use AI to find care?
In the rater8 data, usage was highest in the 45 to 60 age group at 64%, and AI was the leading influence, at 39%, among patients who had actually switched provider. It is not a young-patient behaviour.
How often does AI get practice details wrong?
In the same survey, 66% of the 465 people who had used AI to research providers had met incorrect information, typically addresses, phone numbers, insurance or hours. 60% did not verify what they read anywhere else.
Are these numbers reliable?
They are vendor surveys, self-reported and not peer reviewed, so treat them as direction rather than precision. Two independent surveys pointing the same way is worth acting on. A single figure from either is not worth quoting without the label.
Can I see this traffic in my analytics?
Mostly not. Many assistant answers end without a click, and the visits that do arrive are often attributed poorly. That is why measurement here means asking a fixed set of questions each month rather than watching a referral report.
Sources
- rater8, 2026 Patient Choice Report (April 2026, n=992 US adults, vendor survey): The 47% and 31% usage figures, the influence table, the 45 to 60 and switcher splits, and the 66% and 60% accuracy findings.
- Rock Health, 2025 Consumer Adoption Survey (fielded December 2025, 8,000 US adults, vendor survey): 32% of US adults have used an AI chatbot for health information, double the prior year; 81% took an action afterwards; 64% use it weekly or more.
Published . Last reviewed .
Want this in your business?
Book a call and we will scope exactly which part of your operation an agent should take first.
Book a call with the team