ANNUAL INDUSTRY CONFERENCE
October 15–17, 2026 · Atlanta
See the conference · Get tickets
FOR SPONSORS • EXHIBITORS
October 15–17, 2026 · Atlanta
Request Prospectus • Supporters • Apply
PRACTICE DECISION TOOLS
FOR Doctors, Specialists, NPs, PAs
Calculators • Marketing • CMT Patient Study • All tools
RESEARCH & PROOF
National Stats, Industry Research
Research Briefs • Visuals • Legal Op • Natl Stats
The AI-Informed Patient
One in three American adults now asks a chatbot about their health. Many arrive at your door having already done so, and some never mention it. For a model built on direct physician access, that is either a threat or the strongest value argument you have.
Updated October 2026: evidence grades replaced with evidence types; funding disclosed on every source.
Start here: the questions this brief answers
Tap a question for the short answer, then jump to the evidence.
Roughly a third of U.S. adults have used an AI chatbot for health information, mostly to supplement care, not replace it. A randomized Oxford study found chatbot help did not improve the public's medical decisions. More than one in five AI users have hidden it from their doctor. Meanwhile, trust in physicians and hospitals fell from 71.5% to 40.1% between 2020 and 2024, the trust gap CMT's April 2026 white paper describes. Membership practices, built on access and time, are well placed to become the trusted interpreter: ask every member about AI use, make it safe to share, and offer the human judgment chatbots cannot.
“Your patient already has a second opinion in their pocket. The question isn't whether they'll use it. It's whether they'll bring it to you. Make it safe to say 'I asked a chatbot,' and you've just become the most trusted voice in their health.”
How many patients are using AI for health questions?
About a third. KFF found 32% of adults used an AI chatbot for health information in the past year; Pew found 34% have used one for at least one health purpose.
Go to the full answer ↓Is chatbot health advice accurate?
Mixed at best for the public. In a randomized Oxford trial of about 1,300 people, those using AI chatbots did no better at identifying conditions or next steps than those using normal sources.
Go to the full answer ↓Are patients telling their doctors?
Not always. A survey commissioned by Zocdoc found more than one in five AI users had hidden it from their doctor, while 78% of providers said they want to know.
Go to the full answer ↓Is there a trust gap?
Yes. Trust in physicians and hospitals fell from 71.5% of adults in 2020 to 40.1% in 2024, while 51% of adults in a 2026 survey said they had used AI for an important health decision without consulting a professional.
Go to the full answer ↓Is this a threat to concierge medicine?
It can be either. If AI answers the easy questions, members may question the fee. If the practice becomes the trusted interpreter of AI, access becomes more valuable, not less.
Go to the full answer ↓What should my team actually say and do?
Ask about AI use at every visit, without judgment. Invite members to send what the chatbot said. Correct calmly. The script in Part 5 is a starting point.
Go to the full answer ↓One in three adults, and rising
Three national surveys taken in late 2025 and 2026 tell a consistent story: AI has become a routine health information source, about as common as social media.
Most people use AI alongside care, not instead of it. In the Gallup survey, more than half of AI users said they turned to it before or after seeing a doctor. The most common uses were nutrition or exercise (59%), physical symptoms (58%), medication side effects (46%) and interpreting medical information (44%).3 KFF found younger adults more likely to use AI, especially for mental health.1
The scale is now enormous. OpenAI says more than 230 million people a week ask ChatGPT health or wellness questions, and in January 2026 it launched ChatGPT Health, which can connect to medical records. OpenAI states the tool is "not intended for diagnosis or treatment," and its health lead has said HIPAA does not apply in that consumer setting.4 Industry Research
Strong on exams, weaker in real hands
Large language models pass medical licensing questions. The harder question is whether ordinary people, typing their own symptoms, get better decisions. The first randomized answer is: not yet.
In a study published in Nature Medicine in February 2026, Oxford researchers randomized about 1,300 people to work through physician-written medical scenarios with or without an AI chatbot. Chatbot users did no better than people using their usual sources. Participants often left out key details, models gave different answers to slightly different questions, and responses mixed accurate and inaccurate advice.5 Clinical Trial Evidence
Why the gap exists
- Patients don't know which details matter
- Small wording changes produce different answers
- Correct and incorrect advice arrive in the same confident voice
What it means at the visit
- Expect anchoring on a chatbot's first suggestion
- Expect both false reassurance and false alarm
- Expect better-informed questions, too
The UK data add a warning. In a King's College London analysis of 2,083 adults, 15% had used a chatbot instead of contacting their GP, and 21% of AI users said they decided against seeking care because of something a chatbot said.6 Practice Insight Physicians share the concern: nearly half in the AMA's 2026 survey strongly opposed patients using AI to interpret radiology or pathology results.7
The patients who don't tell you
The provider-and-patient figures above come from a survey commissioned by Zocdoc, a company with a commercial interest in the topic, so CMT labels them Industry Research and treats them as directional.8 The pattern they describe is still plausible and important: patients use AI, assume doctors disapprove, and stay quiet. Silence is the real clinical risk, because the physician can't correct what they never hear.
Why would a member hide AI use from a physician they pay for direct access to? Usually for the same reasons people hide internet searches: fear of seeming foolish, or of wasting the doctor's time. Both are relationship problems, and relationships are what membership practices are built to solve.
Threat, or the strongest value argument you have?
Both readings are defensible. Which one comes true depends on what the practice does next. CMT's analysis below is editorial inference.
The threat case
If a free chatbot answers routine questions at 2 a.m., some members will ask what the membership fee buys. Younger, healthier members may feel the gap most, and they are the members practices most need for long-term panel stability.
The value case
AI produces more information, not more judgment. Members now need someone they trust to sort it, apply it to their history and say "that one doesn't apply to you." That is precisely what membership medicine sells: time, continuity and a physician who answers.
Chatbots make information cheap. They make trusted interpretation more valuable.
The trust gap: what CMT's white paper found
In April 2026, CMT's white paper AI: The Trust Gap Is Already Here brought together independent research showing two lines moving in opposite directions: trust in physicians and hospitals falling, and reliance on AI rising.9 The underlying figures, checked against their original sources:
The trust decline was measured during the COVID-19 pandemic, and the study's authors linked it in part to the politicization of public health; it is not a measure of trust in any one practice or model.10 Patients still want the relationship: in Philips' 2025 Future Health Index, 52% of patients worried about losing the human touch in their care, and patients preferred to get information and reassurance about AI from their doctors.14 Industry Research
“Patients are not embracing AI because they want less humanity in their healthcare. They are using AI precisely because they want more of it.” And: “The physician who understands this is not threatened by the device in the patient's pocket.” “Dismissing the AI tools patients are turning to is not a strategy for rebuilding that trust.”9 For a membership practice, that points to the assets AI can't supply: time, continuity and a physician who answers. This is CMT editorial reasoning, not a research finding.
A practical playbook for the AI-informed member
Ask every time
Add one intake question: "Have you looked anything up, including with AI, since we last talked?" Asking normalizes it.
Make it safe
Invite members to send screenshots of what a chatbot said. Thank them. Never mock the source.
Correct with context
Explain what applies to them and what doesn't, using their history. That is the part AI can't do well.
A sample script for staff and physicians
“A lot of our members check things with AI now, and that's fine. If you ever do, send us what it said. We would rather see it and talk it through than have you worry, or skip something important, on your own.”
Privacy reminder for members
Consumer AI tools generally sit outside HIPAA.4 Members should know that what they type into a chatbot is not protected the way their medical record is. Saying so is part of the advocacy role members pay for.
Know the numbers
A third of adults use AI for health. Assume your members do too, including the ones who don't say so.
Design for disclosure
Build AI questions into intake, portal messages and the annual visit.
Be the interpreter
Position the practice as the trusted second opinion on whatever the internet, or the chatbot, said.
How this brief was built
CMT reviewed nationally representative U.S. surveys published in 2026 (KFF, West Health-Gallup, Pew Research Center), an industry-commissioned survey (Zocdoc, reported by Darwin Research), a UK poll analyzed by King's College London, a randomized trial published in Nature Medicine, and the sources in CMT's April 2026 white paper AI: The Trust Gap Is Already Here, each re-checked against its original publisher. Where the white paper and the original differ, CMT uses the original (KFF reported 56% not confident in telling true from false AI health information). Survey estimates differ because questions and time frames differ; CMT reports each as worded by its source rather than averaging them.
How to read the evidence types
Every key finding is labeled by evidence type. Labels describe the type of evidence, not its value. Each type answers different questions. Funding is disclosed on every source.
Randomized trials and systematic reviews.Best for cause and effect.
Large observational studies and government data.Best for trends at scale.
Surveys, smaller studies and expert consensus.Best for real-world experience.
Company-sponsored or company-reported data that is not peer-reviewed.Best for early signals and operating data.
Statutes, regulation and official guidance.Best for what is required.
What we don't know
- Member-level data: how often members of concierge and membership practices, who tend to be older and have direct physician access, use AI compared with the general public.
- Whether AI use changes visit frequency, adherence or renewal decisions.
- Accuracy of newer health-specific tools that connect to medical records, which launched in 2026 and have not been independently evaluated in published trials that CMT could identify.
How to cite this brief
External review: this brief has not yet been reviewed by an outside expert. When review is complete, the reviewer is credited by name above with any conflicts of interest, and the version number is updated. Reviewers check accuracy and fairness; CMT is responsible for the final content.
Corrections policy: when an error is identified, CMT corrects it in the open and updates the version number above. Send corrections to the editor through conciergemedicinetoday.net.
Related CMT Research Briefs
References
- KFF. Tracking Poll on Health Information and Trust: about a third of adults have turned to AI chatbots for health information. March 25, 2026. Via Connecticut Hospital Association news clip. cthosp.orgFunding: not stated (conducted and published by KFF)
- Pew Research Center. From diagnoses to treatments, why Americans use AI chatbots for health (survey of 3,488 U.S. adults, June 22 to 28, 2026). August 25, 2026. Via Benton Institute. www.benton.orgFunding: not stated (conducted and published by Pew Research Center)
- West Health-Gallup Center on Healthcare in America. One in four U.S. adults use AI for health information (survey of more than 5,500 adults, October to December 2025). Reported by HealthDay, April 15, 2026. www.healthday.comFunding: not stated (conducted by the West Health-Gallup Center on Healthcare in America)
- Medical Economics. OpenAI launches ChatGPT Health, directly linking patient portals to the AI chatbot. January 8, 2026. www.medicaleconomics.comFunding: not stated (news report)
- Bean A, Payne R, et al. Randomized study of large language model assistance for public medical decision-making. Nature Medicine. February 2026. University of Oxford news release, February 10, 2026. www.ox.ac.ukFunding: not stated
- Digital Health. One in seven use AI chatbots instead of seeing a doctor, poll finds (King's College London analysis, 2,083 UK adults, March 2026). May 21, 2026. www.digitalhealth.netFunding: not stated (King's College London analysis, reported by Digital Health)
- American Medical Association. More than 80% of physicians use AI professionally: AMA survey (2026 Physician Survey on Augmented Intelligence). O'Reilly KB. AMA. March 12, 2026. www.ama-assn.orgFunding: not stated (conducted and published by the American Medical Association)
- Darwin Research Group. Nearly one-third of U.S. adults use AI to get health information, surveys show (includes Zocdoc-commissioned survey, February 2026). March 30, 2026. www.darwinresearch.comFunding: industry (Zocdoc, commissioned)
- Concierge Medicine Today Editorial Team. CMT White Paper: AI, the trust gap is already here. Concierge Medicine Today. April 15, 2026. conciergemedicinetoday.orgFunding: Concierge Medicine Today (self-published; no outside funder) CMT original
- Perlis RH, et al. Trust in physicians and hospitals during the COVID-19 pandemic in a 50-state survey of US adults (443,455 adults; 582,634 responses; April 2020 to January 2024). JAMA Network Open. July 31, 2024. Summary via CIDRAP. www.cidrap.umn.eduFunding: not stated
- The Ohio State University Wexner Medical Center. AI in health care survey (SSRS Opinion Panel, 1,007 U.S. adults, January 16 to 20, 2026). April 7, 2026. wexnermedical.osu.eduFunding: not stated (conducted by SSRS for The Ohio State University Wexner Medical Center)
- Becker's Hospital Review. Public comfort with AI in healthcare declines as use rises: survey. April 2026. www.beckershospitalreview.comFunding: not stated (news report)
- KFF. Poll: most who use artificial intelligence doubt AI chatbots provide accurate health information (KFF Health Misinformation Tracking Poll, 2,428 U.S. adults, June 3 to 24, 2024). August 2024. www.kff.orgFunding: not stated (conducted and published by KFF)
- Philips. Building trust in healthcare AI: five key insights from the 2025 Future Health Index. 2025. www.philips.comFunding: industry (conducted and published by Philips)
Educational and informational only. This CMT Research Brief does not constitute medical, legal, tax, financial, accounting or other professional advice, and it does not create a professional relationship of any kind. Statements about laws, regulations, tax rules and payer policies are general, may not reflect the rules in your state, and can change after publication. Consult a qualified attorney, accountant, tax adviser, compliance professional or licensed clinician before acting on anything here.
Independence. Concierge Medicine Today is an independent publication. It does not accept payment for favorable coverage, and it does not favor one practice model over another. Company names and products are mentioned for context only and are not endorsements. Funding is disclosed for every source in the reference list.
Accuracy. CMT verifies figures against their original or best available sources at the time of publication. Where a figure is an estimate, an inference or a company-reported number, the brief says so. This content is not without possible error or omission.
© 2007-2026 Concierge Medicine Today, LLC. All rights reserved.