I Asked ChatGPT to Interview Me Like a Dietitian. Here's What Actually Happened.
Eighteen minutes. That's the average length of a primary care visit, based on EHR data from over 21 million appointments. Eighteen minutes to explain months of bloating, the constipation-diarrhea flip-flopping, the fatigue that shows up two hours after eating, and whatever new thing started last week. I've sat in that chair. I've watched the doctor's hand hover over the keyboard, waiting for me to wrap it up.
So one night I tried something different. Instead of asking ChatGPT a direct question about my gut symptoms, I asked it to interview me, the way a dietitian doing an intake would. Using ChatGPT for gut symptoms this way turned out to be a completely different experience than typing "why am I bloated" into a search bar. This is the actual transcript, what worked, what didn't, and what I'd tell anyone else trying this.
The prompt I used
Here's exactly what I typed, word for word:
"I want you to act like a dietitian doing a first intake session with me. Don't give me advice yet. Just ask me questions, one at a time, to understand my gut symptoms, my history, and what might be going on. Wait for my answer before asking the next question."
That last line matters more than it looks. Most people asking ChatGPT about gut symptoms ask one big question and get one big answer back: a list of possible causes, a list of foods to avoid, maybe a line about seeing a doctor. Useful, but shallow. Framing it as an interview flips the order. It asks before it answers, which is closer to how an actual intake works than how a search engine works.
I wasn't trying to get it to tell me what was wrong with me. I just wanted to see whether being asked good questions, in order, would surface something an eighteen-minute visit wouldn't catch.
What it asked, turn by turn
It started broad: how long had I been dealing with this, what does a typical day of symptoms look like, is there a pattern around meals or time of day. Normal intake stuff. Then it got more specific, and that's where it got interesting.
The questions that were sharp
It asked about sleep quality before it asked about diet, which caught me off guard, since I'd never once heard "how are you sleeping" in a gut health appointment before. It asked whether symptoms were worse in the morning or built up over the day, which turned out to matter for figuring out whether I was dealing with something tied to digestion timing versus something reactive to specific meals.
Someone dealing with SIBO-type bloating that gets worse through the day, better overnight, then resets in the morning would get a very different follow-up path than someone whose bloating is immediate and meal-triggered. A rushed appointment rarely has time to draw that distinction out. The interview did, just by asking questions in a deliberate order instead of jumping straight to "what have you tried."
💡 Worth knowing: If you want sharper follow-up questions, just mention in your prompt that you want to dig into every detail and nuance of your symptoms.
The questions that missed or led somewhere unhelpful
It wasn't all sharp. It asked about food sensitivities early, before it had asked about red flag symptoms like unintentional weight loss, blood, or night waking from pain, the kind of things a gastroenterologist would ask first to rule out something more urgent. It also, at one point, suggested a fairly generic elimination protocol before it had asked enough questions to justify one.
🔎 If you're trying to make sense of a transcript like this on your own: Noorish takes everything you've noticed, tried, and answered across sessions like this one and turns it into a structured gut health action plan, so you're not left holding a long chat log with no next step. Start here →
How I used it
Beyond the one-off interview, this became a recurring thing before appointments. A few ways it actually helped:
Prepping for an appointment. I ran the interview a few days before a follow-up, then copied the whole transcript into a doc. Walking in with "here's what came up when I actually sat down and answered fifteen questions about this" is a different conversation than trying to reconstruct three months of symptoms out loud, on the spot, while someone's already typing notes.
💡 Worth knowing: Save the transcript as you go. Copy and paste it into a note or doc right after, because chat history gets messy fast once you've run this a few times on different days.
Learning the actual vocabulary. Gut health has a lot of specific terms (motility, visceral hypersensitivity, gut-brain axis) that rarely get explained mid-appointment.
Organizing a timeline I could actually use. After a few of these sessions spread across a couple of weeks, I had a rough written record of what I'd noticed and when, instead of a vague sense that "it's been bad lately."
| Use case | What I typed | What came back |
|---|---|---|
| Pre-appointment prep | "Interview me about my symptoms before my GI visit" | A structured Q&A I could paste into notes |
| Vocabulary check | "What does visceral hypersensitivity mean, in plain English?" | A plain-language explanation, no jargon |
| Pattern review | "Here's what I logged this week, any patterns you'd flag?" | A summary I could double check against what I actually felt |
It's essential, however, that any responses are carefully checked by a medical professional. Use it as prep material, never as the final word.
Conclusion
None of this replaces a doctor or a dietitian, and it's not supposed to. What it's actually good for is the part that happens before and after the appointment: getting your own thoughts in order, catching a pattern you'd talked yourself out of noticing, walking in with something more organized than "it's been bad, I don't know, everything maybe." Learning how to use ChatGPT for gut symptoms productively mostly comes down to asking it to ask you questions instead of asking it for answers.
If you're at the stage where you've got scattered notes, half-remembered chat sessions, and a vague sense that something's off but no structure to any of it, that's exactly the gap a tool like Noorish is built for.
🔎 Noorish: Gut Health Action Plan
Turn scattered symptom notes and chat transcripts into an actual action plan.
- ✅ Build a structured gut symptom history to share with your doctor
- ✅ Understand what's actually driving your symptoms
- ✅ Get a science-based action plan for what to try next
- ✅ Optional: validation from a real nutritionist
Follow along on Instagram if you want to see more of this kind of thing tried out in real time.
FAQ
Can ChatGPT tell me if I have IBS or SIBO?
No. It can't order a breath test, run bloodwork, or examine you, all things that are actually part of confirming SIBO or ruling in IBS under the Rome IV criteria (source: Rome Foundation criteria). It can help you organize symptoms and history to bring to someone who can.
Is it safe to use ChatGPT for health advice?
For general education and organizing your own thinking, generally yes. For anything specific to your case, especially red flag symptoms like blood, fever, or unexplained weight loss, no. Those need a real person, quickly, not a chat window.
What should I tell ChatGPT about my symptoms?
Be specific rather than general. Timing (morning vs. evening), triggers you've noticed, what makes it better or worse, and how long it's been going on all produce sharper follow-up questions than a one-line summary like "I have bloating."
How accurate is ChatGPT for gut health questions?
Mixed, and it depends on what you're asking. One study found licensed healthcare professionals preferred chatbot responses over doctor responses to real health questions people posted online, 78.6% of the time, largely on quality and empathy (source: Ayers et al., JAMA Internal Medicine, 2023). But the same researchers were clear that responses still need to be checked by a professional before anyone acts on them.
What are good prompts to ask ChatGPT about my health?
Instead of asking it a direct question, ask it to interview you first: "Ask me questions, one at a time, to understand my symptoms before giving me any advice." It changes the quality of what comes back, because it's building context instead of guessing from one line.
