A family medicine client told us a small story this fall that stuck with us. A long time patient came in for something minor and mentioned, almost as a joke, that she had already asked an AI about it for a week before calling. It told her it was probably nothing. She waited. It turned out to be worth a visit, and she was fine, but the doctor's first thought was not relief. It was a question. How many of my patients are doing this and never calling at all?
That is not a hypothetical anymore. It is measured, and the measurements are striking.
The number: 42% stop at the chatbot
In March 2026, KFF, the health policy nonprofit formerly known as the Kaiser Family Foundation, surveyed 1,343 adults about how they use AI for health. The headline finding, reported by TechTarget, is the one above: 42% of people who used an AI chatbot for a physical health question did not follow up with a healthcare provider afterward. For mental health questions, the share who skipped professional care was even higher, at 58%. The people most likely to stop at the chatbot were younger and lower income, the exact patients a growing practice wants to reach.
Read that again from a practice owner's chair. For four in ten of these people, the AI answer was the final answer. There was no call, no form, no booked slot. The visit simply evaporated, and you never saw a trace of it in your schedule.
And when they do look for a doctor, they ask AI first
The second trend makes the first one worse. Patients are not only using AI to decide whether to seek care. They are using it to decide who to see. The rater8 2026 Patient Choice Report found that the share of patients using AI to research providers jumped from 31% at the end of 2025 to 47% by the middle of 2026. In that same study, AI swayed 36% of patients' care decisions, nudging just ahead of plain Google search at 34%. For the first time, the chatbot outranked the search bar as the thing that tips a patient's choice.
We dug into the full report earlier this year in our breakdown of whether patients are using AI to self diagnose before booking, and the pattern has only accelerated since. The short version: the front door to your practice used to be Google. Now, for a big chunk of patients, there is a step before Google, and it is a conversation with a machine that quietly decides whether you even come up.
The part that should actually bother you
Here is where it gets uncomfortable. That same rater8 report found that two thirds of people who used AI to research a provider ran into wrong information, and 60% trusted the AI summary without checking it. Wrong hours. An old phone number. The wrong insurance listed. An address from two offices ago.
So picture the full chain. A patient asks AI for a good dermatologist nearby. The AI names a few, pulls your details from a listing that has not been touched since 2023, and tells the patient you do not take their plan, when you do. The patient believes it, does not verify, and books the clinic down the street. You did not lose that patient on quality or price. You lost them to a stale line of data you did not know was out there. We wrote a whole piece on this failure mode in whether Google AI recommends your medical practice, and the fix is less about clever marketing than about hygiene.
Our honest take: do not fight the chatbot, surround it
It is tempting to treat this as a threat to argue with. You cannot. You will not out talk a tool that is free, instant, and available at 11pm when a worried patient is lying awake. Telling patients to stop using AI is like telling them to stop Googling in 2010. It is not going to happen, and it makes you sound like the past.
The winning move is the opposite. Be so present, so accurate, and so easy to reach that you are the natural next step the moment the chatbot reaches its limit. And it does reach a limit. KFF also found that most adults still lack full confidence in AI for health, and a majority want a human in the loop. The patient who asked AI is not refusing care. They are gathering the nerve to seek it. If you are the obvious, trusted, frictionless option at that exact moment, you win them. If you are hard to find or slow to answer, the moment passes.
That breaks down into three plain jobs.
1. Be the answer the AI gives
AI tools build their recommendations from the open web and from your business listings. So the work that gets you named by a chatbot is the same work that helps you rank in search: a complete and accurate Google Business Profile, matching name, address, and phone everywhere, steady recent reviews, and a clear website that spells out who you treat, where, and which insurance you take. People call this GEO, generative engine optimization, but it is mostly honest healthcare SEO done consistently. If you want the deeper playbook, we laid it out in how to get your practice recommended by AI. A practice that is invisible or inconsistent online simply gives the model nothing to recommend.
2. Make the next step take seconds, not effort
A patient who leaves the chatbot ready to act will not jump through hoops. If your only path in is a phone call during office hours, you lose every one of them who got the nerve up at 9pm. Real online booking that shows actual open times, a one tap call button, and a simple way to text your practice remove the friction at the exact moment it matters. We make the case for this in why online booking should show real time openings. The clinic that lets a patient grab a slot in three taps beats the one that makes them call back on Monday.
3. Answer first, every time
When a patient finally does reach out, they are deciding in the moment, and the window is short. A reply the next business day is a reply to someone who already booked elsewhere. The research on lead response is brutal and consistent, which is why we keep coming back to it in how fast you should respond to a new patient lead. Minutes matter. For the calls and messages that land after hours or during a packed morning, that is exactly the gap our AI receptionist is built to close, answering, qualifying, and booking the patient while the intent is still hot instead of letting them roll to a voicemail nobody checks until lunch.
Where EtherealMinds fits
We work only with healthcare practices in the United States, and this is the shift we spend most of our time on right now. The old funnel, patient searches, patient finds you, patient calls, has a new, invisible first step, and a lot of practices are bleeding patients there without knowing it. You cannot fix a leak you cannot see.
So we handle the whole chain. We clean up and standardize the listings AI and search pull from, build the reviews and the website that convert and rank, and wire up fast response and after hours coverage through our patient acquisition system so that the patient who started with a chatbot ends up in your chair. The goal is not to beat AI. It is to make your practice the thing AI points to, and the easiest yes a worried patient makes all week.
Curious what AI says about your practice right now?
Book a free strategy call and we will show you what the major AI tools actually say when a patient asks for a provider like you, where your listings are wrong or missing, and how to turn that first chatbot moment into a booked visit. No pressure, just an honest look at your front door.
Book a free strategy call →Frequently asked questions
How many patients use AI instead of seeing a doctor?
In a March 2026 survey of 1,343 adults, KFF found that 42% of people who used an AI chatbot for a physical health question did not follow up with a healthcare provider afterward. For mental health questions the share who skipped professional care was even higher, at 58%. Younger, lower income adults were the most likely to stop at the chatbot. So a large minority of people are now using AI as a substitute for a visit, not a step toward one.
Are patients really using AI to choose a doctor?
Yes, and it is rising fast. The rater8 2026 Patient Choice Report found that use of AI to research providers jumped from 31% at the end of 2025 to 47% by the middle of 2026. In that same report, AI swayed 36% of patients' care decisions, slightly ahead of plain Google search at 34%. The catch is accuracy: two thirds of people who used AI for provider research ran into wrong information, and 60% trusted the AI summary without checking it.
Is a patient using AI a lost patient for my practice?
Not at all. A patient asking AI about a symptom is a patient who is worried enough to look something up, which is a warm lead, not a dead one. KFF found most people still lack full confidence in chatbot answers for health. If your practice shows up clearly in the places AI and search pull from, with accurate hours, phone, and insurance, and a patient can book in a few taps or reach a real voice right away, you become the natural next step after the chatbot rather than the visit that never happened.
How do I get my practice mentioned when a patient asks AI for a doctor?
AI tools lean on the open web and on your business listings to decide which provider to name. That means the fundamentals still win: a complete and accurate Google Business Profile, consistent name, address, and phone across directories, real reviews, and a fast, well structured website that clearly states who you treat and where. This overlap of search and AI is often called GEO, and it is the same authority work that helps you rank in Google. A practice that is invisible or inconsistent online is a practice the AI has no reason to recommend.
What is the single fastest fix to capture these patients?
Speed of response. A patient who leaves the chatbot and finally reaches out is deciding in the moment, and that window is short. Make sure every inquiry, call, form, or text, gets answered within minutes, not the next business day. Online booking that shows real openings and an AI receptionist that answers after hours both close that gap. The practice that responds first usually wins the patient, no matter who had the better reviews.