Fewer Patients | Deeper Insights | Better Outcomes | Limited Memberships | Fewer Patients | Deeper Insights | Better Outcomes | Limited Memberships | Fewer Patients | Deeper Insights | Better Outcomes | Limited Memberships | Fewer Patients | Deeper Insights | Better Outcomes | Limited Memberships | Fewer Patients | Deeper Insights | Better Outcomes | Limited Memberships | Fewer Patients | Deeper Insights | Better Outcomes | Limited Memberships | Fewer Patients | Deeper Insights | Better Outcomes | Limited Memberships | Fewer Patients | Deeper Insights | Better Outcomes | Limited Memberships | Fewer Patients | Deeper Insights | Better Outcomes | Limited Memberships | Fewer Patients | Deeper Insights | Better Outcomes | Limited Memberships | Fewer Patients | Deeper Insights | Better Outcomes | Limited Memberships | Fewer Patients | Deeper Insights | Better Outcomes | Limited Memberships | Fewer Patients | Deeper Insights | Better Outcomes | Limited Memberships | Fewer Patients | Deeper Insights | Better Outcomes | Limited Memberships | Fewer Patients | Deeper Insights | Better Outcomes | Limited Memberships | Fewer Patients | Deeper Insights | Better Outcomes | Limited Memberships | Fewer Patients | Deeper Insights | Better Outcomes | Limited Memberships | Fewer Patients | Deeper Insights | Better Outcomes | Limited Memberships | Fewer Patients | Deeper Insights | Better Outcomes | Limited Memberships | Fewer Patients | Deeper Insights | Better Outcomes | Limited Memberships |
Dr. Sokol's leather doctor's bag with a stethoscope and supplies for concierge house calls
11/05/2026

Can You Trust AI With Your Health? Even Doctors Need to Be Careful

Why artificial intelligence in healthcare makes decades of clinical experience more important, not less.

AI tools like ChatGPT can help you research health questions, but they can deliver incomplete or incorrect answers with complete confidence. They are useful for research and for preparing questions. They should not replace a physician who can evaluate the evidence against your history and take responsibility for the decision.

According to OpenAI, the company behind ChatGPT, artificial intelligence can produce answers that sound authoritative even when they are incorrect:

“Sometimes, ChatGPT sounds convincing, but it might give you incorrect or misleading information.”

OpenAI, in its guidance on ChatGPT’s accuracy

The World Health Organization has issued a similar warning about large language models (LLMs), the technology behind AI tools such as ChatGPT:

“These responses may be completely incorrect or contain serious errors, especially for health-related responses.”

World Health Organization, May 2023

Dr. Matthew Sokol, MD, FMCP-M, is a board-certified medical doctor with more than two decades of clinical experience and advanced certification through the Institute for Functional Medicine. Recognized multiple times among D Magazine’s Best Doctors in Dallas, he has spent his career evaluating patients, interpreting clinical findings, making treatment decisions, and observing outcomes over time.

His perspective on artificial intelligence is grounded in a fundamental responsibility: a physician must be able to independently evaluate the information used to make decisions about a patient’s health.

“I’ve spent time studying what artificial intelligence can and cannot do in medicine, and I think it has tremendous potential. But I’ve also seen how convincingly these systems can present information that is incomplete, misleading, or simply wrong. That concerns me, particularly when people begin treating the output as a clinical conclusion.

I believe AI can be useful for research and certain administrative efficiencies, but I draw a firm line at allowing it to substitute for my own clinical judgment. I need to understand the evidence, evaluate it against the patient’s history and findings, and be able to stand behind every decision I make.

After more than two decades of practicing medicine, I know that some of the most important clinical decisions come from recognizing what doesn’t fit. Technology can help us access information. It cannot take responsibility for the patient sitting in front of me.”

Dr. Matthew Sokol, MD, FMCP-M, Board-Certified Medical Doctor

What happens when a medical recommendation sounds completely credible, references published research, and reaches the wrong conclusion?

Now consider a more uncomfortable possibility.

What happens when the person relying on that recommendation is your doctor?

Artificial intelligence is changing medicine at extraordinary speed. Patients use ChatGPT to research symptoms, interpret bloodwork, and investigate possible diagnoses. Physicians are increasingly using AI to review medical literature, summarize clinical information, and assist with their work.

The potential is enormous.

But there is a distinction that deserves considerably more attention: using artificial intelligence to support medical judgment is very different from relying on it to supply that judgment.

And as AI becomes more sophisticated, that distinction may become harder to recognize.

AI’s Confidence Problem: When the Answer Sounds Right but Isn’t

One of the most misunderstood characteristics of artificial intelligence is its ability to communicate with extraordinary confidence.

Ask ChatGPT to explain an elevated inflammatory marker or an abnormal thyroid result, and you may receive a polished response with detailed medical terminology, plausible explanations, and suggestions for further investigation.

The answer may be useful.

It may also be incomplete, misleading, or incorrect.

AI systems can generate what researchers call hallucinations: information that appears factual but is inaccurate or invented.

In medicine, that can mean citing research that doesn’t exist, misrepresenting findings from a legitimate study, overlooking an important clinical detail, or recommending an intervention that isn’t appropriate for the patient.

And because the language is often so convincing, recognizing the error can be difficult.

The World Health Organization has specifically warned that large language models can produce authoritative-sounding medical information containing serious errors.

OpenAI itself acknowledges that ChatGPT can generate incorrect information while sounding confident.

The technology can be impressive and still require someone capable of recognizing when it is wrong.

Even Physicians Are Increasingly Using AI

Artificial intelligence is no longer something happening outside the doctor’s office.

According to the American Medical Association’s 2026 Physician Survey on Augmented Intelligence, 81% of surveyed physicians reported using AI professionally, more than double the percentage reported in 2023.

Among the most common applications were summarizing medical research, reviewing standards of care, and assisting with clinical documentation. Some physicians also reported using AI for diagnostic assistance.

These developments offer meaningful opportunities.

AI may help physicians navigate an expanding body of medical literature, reduce administrative demands, and identify information that warrants closer examination.

But greater adoption introduces an important responsibility.

A physician must be able to independently evaluate the information AI produces, particularly when it may influence patient care.

A sophisticated recommendation is not automatically a reliable recommendation.

A medical study cited by AI is not necessarily an accurately represented study.

And a diagnosis suggested by an algorithm is not a diagnosis until the clinical evidence supports it.

The concern isn’t that physicians are using AI.

It’s what happens when they begin trusting its conclusions without sufficiently questioning them.

AI-Assisted Medicine vs. AI-Dependent Medicine

There is an important difference between using technology to improve access to information and allowing technology to determine the direction of care.

An AI-assisted physician may use a tool to identify relevant research, explore possible explanations, or organize information for further review.

The physician then evaluates that information independently.

Does the evidence support the conclusion? Is the research reliable? Does the recommendation account for the patient’s medical history, medications, examination findings, and other relevant factors?

An AI-dependent approach risks allowing the generated answer to become the starting assumption rather than a claim that must be evaluated.

That distinction matters because AI systems can produce conclusions that are plausible but clinically inappropriate.

Consider a patient with fatigue, changes in weight, and abnormal thyroid measurements.

An AI system may suggest a particular thyroid disorder as a likely explanation.

But an experienced physician might recognize that a medication, recent illness, or another medical condition changes how those results should be interpreted.

Or an AI-generated summary might recommend an intervention based on a study whose findings are preliminary, limited, or not applicable to the individual patient.

The physician needs to recognize those limitations.

Sometimes the most important clinical decision is not accepting a recommendation, but challenging it.

The value of an experienced physician isn’t diminished by having access to more information. It’s demonstrated by knowing what to trust, what to question, and what to disregard.

For patients who want that level of scrutiny applied to their own care, a complimentary discovery call with Dr. Sokol is a chance to explore whether the approach fits.

The Clinical Learning Curve Cannot Be Automated

Medical knowledge and clinical judgment are closely related, but they are not interchangeable.

A physician can retrieve information about a condition in seconds.

Understanding how that information applies to a particular patient requires another level of expertise.

Clinical judgment develops through medical education, training, and decades of caring for patients.

Over decades of practice, physicians encounter conditions that present atypically, patients who respond unexpectedly to treatment, and situations where the most obvious explanation turns out to be incomplete.

They learn to recognize patterns, but also to recognize when a pattern is misleading.

They develop an understanding of how medications, medical histories, symptoms, examination findings, and diagnostic results interact.

And they observe what happens after a treatment decision is made.

Did the intervention help? Did new symptoms emerge? Did the patient’s response support the original assessment, or did it suggest a different explanation?

These experiences inform future decisions.

Dr. Sokol’s additional functional medicine training expands the range of factors he considers when evaluating a patient’s health, while his extensive medical experience provides an essential foundation for interpreting findings and determining appropriate treatment.

The combination matters.

Comprehensive medicine should not mean pursuing every available test, accepting every emerging theory, or treating every abnormal result.

It should mean applying thoughtful medical judgment to a more complete understanding of the patient.

Medical Research Requires Interpretation, Too

There is another limitation of AI that receives less attention.

Finding a medical study is not the same as understanding what the study establishes.

Research varies in quality, methodology, sample size, and clinical relevance.

A finding may be statistically significant without being clinically meaningful.

A treatment may show promise in a small preliminary study without sufficient evidence to support widespread use.

And results observed in one population may not apply to a patient with different medical circumstances.

AI can help identify and summarize research, but its interpretation still needs to be checked against the original evidence.

A physician must understand not only what a study reports, but whether its findings justify a particular clinical decision.

That is especially important in functional and integrative medicine, where patients may encounter emerging research, specialized testing, and treatments supported by varying levels of evidence.

The objective should be thoughtful investigation grounded in appropriate clinical evidence.

The Physician-Patient Relationship Cannot Be Automated

Technology can make medical information more accessible.

It can help organize records, identify possible drug interactions, and support clinical research.

But medicine involves more than processing information.

It requires understanding the person behind the data.

Two patients with similar symptoms may have very different medical histories, risk factors, treatment priorities, and appropriate courses of care.

A physician who knows a patient over time has the opportunity to understand those differences and evaluate how the patient’s health evolves.

That continuity becomes particularly important when a treatment doesn’t produce the expected result.

An AI-generated recommendation can be revised with another prompt.

A physician must decide whether to continue treatment, investigate further, change direction, or recognize that an earlier assumption was incorrect.

Those decisions carry real consequences.

What Patients Should Look for in an AI-Driven Healthcare World

As artificial intelligence becomes more integrated into healthcare, patients have another important consideration when choosing a physician.

Not simply whether the physician uses modern technology.

But whether the physician has the experience and judgment to use it responsibly.

For patients exploring concierge medicine or functional medicine in Dallas, longer appointments and more direct physician access are meaningful advantages.

Yet those features matter most when paired with clinical expertise and continuity.

Dr. Sokol’s approach combines more than two decades of medical experience as a board-certified medical doctor with advanced functional medicine training and a longitudinal model of care.

The Ivalo MD Blueprint includes a comprehensive medical history, physical examination, and individualized diagnostic evaluation. Findings inform a physician-written treatment plan, with ongoing Clinical Strategy Sessions to assess progress and make adjustments when appropriate.

The practice maintains an intentionally limited patient panel to preserve that direct physician relationship.

Located in McKinney at the Prosper line, Ivalo MD serves patients throughout Dallas, Plano, Frisco, Celina, and the greater Dallas–Fort Worth area.

The Future of Medicine Needs Both Innovation and Judgment

Artificial intelligence will continue to influence how medical information is accessed, organized, and interpreted.

That presents real opportunities for patients and physicians alike.

But the sophistication of a technology should never be confused with the reliability of every conclusion it produces.

The right question is not whether AI belongs in medicine.

It is whether the person using it has the expertise to recognize its limitations and take responsibility for the decisions that follow.

Because in healthcare, knowing how to find an answer matters. Knowing when that answer is wrong matters even more.

Explore Ivalo MD Memberships

Or begin with a complimentary discovery call with Dr. Sokol.

Frequently Asked Questions

Is ChatGPT accurate for medical advice?

It can be useful for research, but both OpenAI and the World Health Organization warn that it can produce incorrect information that sounds authoritative.

Do doctors use AI?

Yes. In the American Medical Association’s 2026 survey, 81% of physicians reported using AI professionally, more than double the 2023 figure.

What is an AI hallucination?

Information that appears factual but is inaccurate or invented, such as a citation to research that does not exist.

How does Dr. Sokol use AI?

For research and certain administrative efficiencies, never as a substitute for his own clinical judgment.

Should I upload my lab results to ChatGPT?

It can help explain what a marker measures, but it cannot examine you, know your full history, or take responsibility for a treatment decision.

Join a select group of patients who have discovered what's possible when healthcare is done right.

At Ivalo MD, we’ve broken away from the status quo. Because healthcare shouldn't be something you have to fight for. We've created a concierge functional medicine practice that puts you first—where proactive, personalized care is the norm, not the exception. Where your doctor knows you by name, not just by chart number. And where your health goals aren't just heard, they're actively pursued.

Ready to experience healthcare as it should be?

EXPERTISE SO RARE, IT CHANGES EVERYTHING