You leave the doctor's office with a printout, a test result, or a prescription โ€” and within a few hours you're staring at a screen, trying to make sense of words you've never heard before. Or your parent gets a diagnosis and you want to understand what it actually means before the follow-up appointment. Or you're reading through a pathology report and there are numbers with ranges printed next to them and you don't know if the ones outside the range are serious or routine.

Most people in this situation end up on one of two paths: they ask Google, which leads them into a spiral of worst-case scenarios and forum posts from 2011, or they wait until the next appointment to ask questions โ€” by which point they've been anxious for a week and can't remember half of what they wanted to know.

AI tools like Claude and ChatGPT offer a third path that most people haven't fully explored yet. Used the right way, they can help you understand what you're looking at, prepare better questions for your doctor, and get out of the anxiety loop that comes from not knowing what you don't know. Used the wrong way, they can send you in circles or โ€” worse โ€” give you confident-sounding information that doesn't apply to your actual situation.

The difference between those two outcomes is mostly about how you use the tool. That's what this article is about.

What AI Is Actually Good At Here

Before getting into specific approaches, it's worth being clear about what AI tools are and aren't suited for in a medical context. This matters, because the line is genuinely important.

AI is good at explaining what things mean in plain language. Medical terminology is dense and often deliberately technical. When a lab report says your "eGFR is 58 mL/min/1.73mยฒ" and you have no idea what that means, Claude or ChatGPT can explain what eGFR measures, what the number indicates, and what the reference range generally means for people in different situations. It is doing what a very knowledgeable friend would do โ€” translating jargon into something you can actually work with.

AI is also good at explaining how conditions and treatments work in general terms. If your doctor mentions a procedure, a medication class, or a diagnosis you've never heard of, the AI can give you a solid overview of what that thing is โ€” how it works, what it's used for, what the general treatment landscape looks like. This is background education, not personal medical advice.

And AI is good at helping you generate questions. This is one of the most underrated uses. Many people leave doctor's appointments without asking the questions they should have asked โ€” not because they didn't have them, but because they didn't know what to ask. AI can help you work out what you don't yet understand and turn that into a list of specific, useful questions to bring to the next appointment.

What AI is not good at: interpreting your specific results in the context of your full medical history, making a diagnosis, or telling you what to do. A language model is not examining you, does not know your history, and can generate plausible-sounding explanations that don't fit your actual situation. The tool is a translator and educator, not a clinician.

How to Ask an AI About a Medical Term or Result

The most common mistake people make when asking AI about medical information is asking the wrong kind of question. "What does this mean for me?" is a question the AI cannot reliably answer. "What does this term mean?" is a question it handles well.

Here's the general approach that tends to work. When you encounter a term, a number, or a phrase you don't understand, try something like:

"Can you explain what [term] means in plain language? I'm not a medical professional. I'm trying to understand what I'm looking at on a lab report."

The second sentence matters. Telling the AI you want a plain-language explanation, rather than a technical one, consistently produces more useful output. Without that framing, you can get a response that's technically accurate but still full of terms you need to look up.

For lab results with reference ranges, you can provide the actual numbers and ask what the ranges typically indicate โ€” while being explicit that you understand the AI cannot interpret your specific results:

"My TSH result is 6.2, and the reference range on the report shows 0.5โ€“4.5. Can you explain what TSH measures and what it generally means when the result is above the reference range? I know you can't tell me what this means for my specific situation โ€” I just want to understand the concept before my next appointment."

That kind of framing produces better responses because it's honest about what you're asking for. You're not asking the AI to be your doctor. You're asking it to help you understand a concept well enough to have a more informed conversation with your actual doctor.

Using AI to Understand a Diagnosis

When someone you care about โ€” or you yourself โ€” receives a diagnosis, the period immediately after is often a fog. You hear the name of the condition, maybe a few things about it, and then a lot of logistical information about next steps. By the time you get home, you realize you still don't quite understand what the diagnosis actually is or what it means in practical terms.

AI is genuinely helpful for this kind of orientation. A prompt that tends to work well:

"My [family member] has just been diagnosed with [condition]. Can you give me a plain-language overview of what this condition is โ€” how it works, what it generally affects, and what the typical treatment approaches look like? I'm not looking for specific advice about their case. I just want to understand the basics so I can be more useful during appointments and conversations with their care team."

The "I'm not looking for specific advice" framing is worth including. It tends to produce a more straightforwardly educational response and reduces the chances of the AI hedging its way through every sentence with disclaimers that don't add useful information.

One useful follow-up once you've got the overview: ask what questions a patient or family member would typically want to ask their doctor about this condition. AI is good at surfacing the questions you didn't know you should be asking โ€” the ones about prognosis, about lifestyle, about what to watch for, about how the treatment affects daily life. That list becomes the foundation of your next appointment conversation.

Making Sense of a Medication

Prescription information sheets are written for regulatory compliance, not for human comprehension. They list every possible side effect the drug has ever been associated with, in order of frequency, in dense paragraphs, without any indication of which ones are common and which ones affect one person in fifty thousand. Reading them in full is not always useful. Sometimes it produces unnecessary anxiety about things that are vanishingly unlikely.

AI can help you get a more proportionate picture. A useful framing:

"I've just been prescribed [medication name]. Can you explain what it's typically used for, how it generally works, and what the most commonly experienced side effects tend to be โ€” as opposed to the rare ones that appear on the package insert? I'd also like to know if there are common things patients are usually advised to watch for or avoid while taking it."

This gives you a proportionate, practical orientation rather than the anxiety-maximizing experience of reading the full insert. You still read the insert โ€” that information is there for a reason โ€” but you approach it with context rather than without it.

One caveat worth stating clearly: drug interactions are an area where you should lean heavily on your pharmacist and doctor rather than AI. The AI's knowledge has a training cutoff and may not reflect the most current guidance on specific interaction risks. Your pharmacist has access to your complete medication list and is trained specifically to catch these. Use AI for orientation; use your pharmacist for verification.

Preparing Questions for Your Doctor

This is where AI earns its keep in the medical context, and it's worth spending a moment on it specifically. Most people do not leave medical appointments feeling like they fully understood what was said and asked everything they needed to ask. The appointment is short. The doctor is moving. There's a lot of new information arriving at once, and you often don't know what you don't know until you're in the car on the way home.

Spending fifteen minutes with Claude or ChatGPT before a significant appointment can genuinely change that experience. You describe the situation, what you've been told so far, and what you're going in to discuss, and you ask the AI to help you think through what questions would be most useful to bring.

"I have an appointment next week with a cardiologist. I was referred after my GP noticed an irregular heartbeat on a routine ECG. I don't have any symptoms and I don't know much about what comes next. Can you help me put together a list of questions I should be thinking about asking โ€” things about what tests they might recommend, what the possible explanations are, what the monitoring or treatment options might look like, and what I should be paying attention to in the meantime?"

The list you get back won't be exhaustive, and some questions may not apply to your situation. But it gives you a starting point that's much better than a blank notebook. Review it, add the things that feel relevant, remove the ones that don't, and bring it in. Writing down questions before an appointment โ€” and looking at them during the appointment โ€” is one of the most consistently useful things you can do as a patient. AI just makes generating that list faster.

A Few Things Worth Keeping in Mind

AI tools can give confident-sounding answers that turn out to be wrong or outdated. This is a known limitation โ€” sometimes called hallucination, where the AI states something with apparent certainty that isn't accurate. In a medical context, this matters more than most. Use AI to get oriented and generate questions, but verify anything important through your care team or reliable sources like the NHS website, the Mayo Clinic, or equivalent reputable health information organizations.

Claude tends to be slightly more cautious and careful with medical information than ChatGPT, in my experience โ€” it's more likely to flag uncertainty clearly and recommend professional consultation. That's worth knowing when you're choosing which tool to use for this kind of task. Either works; Claude just tends to hedge a little more explicitly, which in this context is probably a feature rather than a bug.

And finally: the goal of using AI in this context is not to arrive at the appointment already knowing the answer. It's to arrive better prepared to have a real conversation. Your doctor knows your history, can examine you, and can exercise clinical judgment that no AI tool can replicate. The AI is helping you be a more informed participant in that conversation. That's a meaningful thing. It's also exactly the right boundary for where its role should stop.