Asked ChatGPT about your symptoms? Here's what a doctor wants you to know

You've had a headache for a few days. Nothing terrible, but enough to make you wonder what's going on. So you open ChatGPT, or another AI chat tool, describe your symptoms, and ask what might be causing them.

Within seconds, you have a list of possible conditions, suggested tests, and maybe even a treatment plan. But how much of that information should you trust? And what happens when you bring it to your doctor?

For Brian Leonard, DO, a family medicine physician at Quail Creek Clinic for Intermountain Health, these conversations are becoming increasingly common. On a typical day, one to three patients mention using AI to research their health. And it's not necessarily the younger crowd.

According to Dr. Leonard, it's not necessarily younger patients who are turning to AI for health information. In fact, he's finding that many Gen Y and Gen Z patients leave the research to their doctors, while baby boomers are often the ones arriving at appointments having already consulted AI.

His reaction? He welcomes it.

"I love it," says Dr. Leonard, who sees AI as another way for patients to become more informed and involved in their health.

But there's a difference between researching your symptoms and diagnosing yourself. Knowing that difference can help you use AI without putting your health at risk.

What is AI chat, and how does it answer health questions?

ChatGPT, Claude, Gemini, and other AI chat tools are a type of artificial intelligence called a large language model, or LLM. Think of it as a powerful digital assistant trained to recognize patterns, understand questions, and generate useful responses.

Unlike a traditional search engine that gives you a list of websites, these tools can pull ideas together, explain complicated information in everyday language, and have a back-and-forth conversation with you.

It works by learning patterns from huge amounts of information and using those patterns to build responses, one piece at a time. Some AI tools can also search the internet for updated information.

But there's an important catch: AI doesn't automatically know whether everything it tells you is true. It can misunderstand your question, miss important details, or give an answer that sounds convincing but is wrong.

And when the question involves your health, those mistakes can have real consequences.

Can AI tools accurately diagnose your symptoms?

Dr. Leonard compares using AI for health questions to trying to figure out what's wrong with your car.

You hear an unusual noise, look it up online, and decide it might be the timing belt. When you take your car to the mechanic, you may be on the right track. But the mechanic still needs to examine the engine to find the problem.

The same idea applies to healthcare.

AI can help you understand symptoms, explain unfamiliar medical terms, and identify questions worth asking. But it doesn't have the full picture of your health.

Your doctor considers your medical history, medications, family history, physical exam, and other details that may change what a symptom means.

Dr. Leonard has cared for many of his patients for more than a decade. In some cases, he also treats their family members, giving him valuable insight into their health history and potential risks.

AI chat tools can only work with the information available to them. And even with plenty of details, they can reach the wrong conclusion.

Why AI self-diagnosis can lead to unnecessary worry and testing

One risk of AI self-diagnosis is that it can point people toward unlikely conditions and unnecessary testing.

Dr. Leonard recalls a patient who brought in what appeared to be emergency room paperwork recommending tests for several autoimmune conditions. The document looked so convincing that he initially thought it came from a hospital. In reality, the patient and family used AI to create it based on their symptoms.

AI can also make routine results seem alarming. Someone with mild kidney disease, for example, might worry they're headed for dialysis after entering their lab results into ChatGPT, even though their doctor sees no immediate cause for concern.

A possible diagnosis isn't always a likely one. And AI doesn't always make that distinction clear.

How to ask AI chat tools better questions about your health

The way you phrase a question can influence the answer you receive.

Dr. Leonard recommends giving AI a role before asking your question. For example, you might start with: "Act as a medical educator with decades of experience helping me understand a health concern."

Then explain what you want to learn. You can ask it to describe common causes, explain medical terms, or suggest questions for your doctor. Giving AI a role can help shape a more focused response, although it doesn't make the tool a qualified medical professional.

Just as important, avoid steering AI toward a diagnosis you already suspect.

Instead of asking, "Could my headache be a brain tumor?" try, "What are common causes of headaches, and which symptoms need medical attention?"

Or, instead of "Does this blood test mean I have kidney disease?" ask, "What does this blood test measure, and what factors can affect the results?"

Dr. Leonard also encourages people to challenge AI-generated answers. Ask where the information came from, request reliable medical sources, and check whether the answer changes when you ask a question differently.

Even then, don't assume an answer is correct just because AI provides a source. Look at the original information yourself.

The do's and don'ts of using AI for medical advice

AI can be a helpful research tool, but a few boundaries make it safer to use.

Do:

• Use AI to learn about general health topics and medical terms.
• Give it a role, such as a health educator, to help guide its response.
• Ask open-ended questions rather than leading it toward a diagnosis.
• Check medical claims against trusted healthcare sources.
• Use AI to prepare questions for your doctor.

Don't:

• Copy and paste your entire medical record into a public AI tool.
• Share identifying information or sensitive health details without understanding its privacy policies.
• Assume an AI-generated diagnosis or treatment recommendation is correct.
• Start, stop, or change medications based on AI advice.
• Delay medical care while trying to get an answer from AI.

Dr. Leonard is especially cautious about privacy. Even when patients want help understanding their medical records, he encourages them to avoid sharing identifying details.

And some symptoms shouldn't involve an AI search at all. Chest pain, trouble breathing, or sudden severe symptoms may require emergency care. Don't use AI chat to decide whether an emergency can wait.

Should you tell your doctor you used an AI chat tool?

Absolutely. Dr. Leonard would rather know what patients have been reading than have them worry about a diagnosis or question a treatment plan.

Sometimes AI raises an idea worth exploring. Other times, a physician can explain why a certain diagnosis doesn't fit or why additional testing might cause more harm than benefit.

Dr. Leonard sums up his approach with a phrase he's heard from a colleague: "New tools, old rules."

Healthcare technology continues to change, but the need to evaluate information carefully remains.

At Intermountain Health, primary care providers work with patients to understand their concerns, review their health history, and make informed decisions together. Whether your questions come from ChatGPT, Google AI search, an online article, or something you've noticed about your body, they're worth discussing.

So go ahead and ask questions. Stay curious. And bring what you've learned to your next appointment. Your research can help start the conversation, and your doctor can help you decide what it means for your health.