A new patient in her 40s recently walked into one of our clinics with a printout of a ChatGPT conversation. For three days, she had been typing her symptoms into the chatbot, describing chest tightness and a fatigue that was not going away, asking whether she should be worried.
The reason she turned to AI was straightforward. She had tried to schedule a primary care appointment with her previous provider and was told the next available slot was six weeks out. So, she did what an increasing number of Americans now do when the system says “no.” She opened a browser and asked an AI chatbot.
Her story is no longer unusual. AI conversations are quietly becoming the front door to American health care, and clinicians are seeing the printouts arrive every day. The most important question is not whether ChatGPT is right.
Instead, what we should be asking is: Why are so many patients turning to AI before reaching out to us? The rise of AI self-diagnosis is not a technology story. It is an access story, and clinicians need to read it that way.
A measurable shift in how patients seek care
This pattern is no longer anecdotal. Roughly one in six U.S. adults now report using AI chatbots for health information at least once a month, with the highest rates among adults under 30, according to KFF’s Health Misinformation Tracking Poll. Similarly, Pew Research Center has documented a steady rise in the share of Americans turning to generative AI for medical symptoms and treatment options, and Google Trends data shows search volume for “ChatGPT symptoms” and “AI diagnosis” climbing since 2023.
The trust patients place in these tools is not unfounded. A widely cited study in JAMA Internal Medicine found that ChatGPT responses to patient questions were rated higher in quality and empathy than physician responses in 79 percent of evaluated cases.
Additional research has shown that large language models can match or exceed clinicians on diagnostic vignettes, though performance drops sharply when patients describe symptoms in incomplete or atypical ways. That, of course, is exactly how real patients communicate.
Why patients reach for AI, and what AI gets wrong
It is tempting to read these numbers as a referendum on physicians. They are not. They are a referendum on access.
Consider that the average wait time to see a family medicine physician is 23.5 days, up by 14 percent since 2022, according to AMN Healthcare. The Association of American Medical Colleges projects a shortfall of up to 40,400 primary care physicians by 2036. The National Association of Community Health Centers estimates that roughly 100 million Americans lack a regular source of primary care. The discouraging statistics don’t end there: The Commonwealth Fund ranks the United States last among high-income countries on access to care, and KFF finds that one-third of American adults skipped or delayed care in the past year because of cost.
When the system makes a 30-second AI conversation easier than a 30-day wait, patients will choose AI every time.
The clinical reality of AI self-diagnosis is more nuanced than either its evangelists or its critics admit. Used well, AI can explain a confusing lab result, walk through medication side effects, and help patients articulate symptoms more clearly before visits. Some chatbots now route red-flag symptoms toward emergency care, and for low-acuity mental health concerns.
However, the harms are real and, often, clinical. AI can miss atypical presentations of myocardial infarction in women, sepsis in older adults, and stroke in younger patients. It can also offer false reassurance.
For example, a recent study in the Journal of Medical Internet Research found that patients were more likely to delay care after receiving a benign-sounding AI explanation. Models have hallucinated medications and dosages in peer-reviewed clinical evaluations. They cannot palpate an abdomen or listen to a heart murmur. Performance is consistently worse on health questions from non-English speakers and under-represented populations, the same groups already facing the worst access barriers.
The real prescription is access
AI self-diagnosis is a symptom. The disease is access. If we want patients to use the health care system the way it was intended, we have to make it usable.
That requires three commitments from health care leaders. First, treat access as a clinical quality measure. A diagnosis delayed by four weeks is a diagnosis worsened, and our metrics should reflect that.
Second, invest in same-day primary care models. Walk-in availability, extended hours, integrated labs and imaging, and care delivered across clinics, virtual visits, and the home close the gap that AI is currently filling.
Third, build clinician-AI collaboration rather than patient-AI substitution. AI works best when it supports a clinical relationship, not when it stands in for one.
The honest truth is that patients will keep using ChatGPT. Our job is not to compete with it. Our job is to make sure that when AI says, “You should see a doctor,” there is actually a doctor available to see.
The patient I described at the top of this piece came to one of our clinics that morning. She was evaluated within 20 minutes and left with a real diagnosis, a real plan, and a real human who knew her name. AI did not fail her. The system did. And until we fix the front door, ChatGPT will continue to be the receptionist America never hired.
Payam Zamani is the founder and chief executive officer of MY DR NOW, one of the nation’s fastest-growing independently owned primary care organizations. What began as a single clinic during his medical residency has grown into the Southwest’s largest privately owned primary care group, with more than 75 locations across Arizona and Texas and a team of more than 1,000.
His vision for MY DR NOW was shaped by watching his single mother struggle to find timely care for her family. Dr. Zamani built the organization to fit real life, offering walk-in and scheduled visits, extended hours, telehealth, and in-home house calls under one roof. The practice provides family medicine, pediatrics, women’s health, and chronic disease management, anchored in long-term patient relationships.
Headquartered in Chandler, Arizona, MY DR NOW is on track to become the country’s largest independently owned primary care group under a single brand by 2027. Dr. Zamani is affiliated with Ross University School of Medicine, and his work has been featured in Pulse 2.0, Authority Magazine, citybiz, and Empowered Patient Radio. He shares updates on LinkedIn, with company updates available through MY DR NOW on LinkedIn.
