Should You Be Getting Your Health Advice From AI?

Be honest. Have you done this? 

You've typed a symptom, a supplement question, or a "why am I so tired all the time" into ChatGPT or Googled something and clicked on the AI summary at the top. I think most of us have!

More than 1 in 5 adults worldwide are now turning to AI chatbots for health questions. It's free, it's instant, and it doesn't make you feel judged for asking something you'd feel a bit embarrassed raising with your GP, so I get the appeal completely.

But as a nutritionist, I want to be honest about this, because when it comes to health and nutrition specifically, AI has some limitations that I don't think are being talked about enough, and some of them could genuinely affect your health.

🤖 Where AI is actually useful

Let's be fair first. AI isn't all bad for health information, as it can be super helpful for a few things.

If you've come out of a GP appointment with a term you didn't fully understand, or you've been given a blood test result that means nothing to you, asking an AI to explain it in plain English is a reasonable starting point. It can translate medical jargon quickly and accessibly, and it can also help you think of questions to ask your doctor before an appointment, which I actually think is a brilliant use for it.

And for general, well established information, it can be a decent overview. The broad strokes of what protein does, or why sleep matters, are unlikely to be wildly wrong.

The problems start when you move beyond the general, which is exactly what many people do.

⚠️ The part you get no warning about

Nearly half of the answers provided by leading AI chatbots to common health questions contain misleading or problematic information, according to a study published in BMJ Open (you can click here to read the study). That is not a small margin of error... That is almost a coin flip.

AI tools can't examine you or run tests the way a healthcare professional can, and they don't have the ability to reason through how they came to a conclusion, which is necessary for making safe and accurate medical decisions.

Then there is something called hallucination, which sounds oddly poetic but is anything but. Hallucination is when AI provides a confident, plausible-sounding response that is based on nothing. It has made up fake citations, fake research studies, and fake medical details, all delivered in the same calm, authoritative tone as the accurate information. You would have no way of knowing the difference!

One study slipped fabricated medical details, including made-up lab tests and invented conditions, into questions put to different AI chatbots. Rather than catching the false information, the models created confident explanations for things that don't even exist.

This is extremely alarming when you think about the context. 

Someone is already worried about their health. They ask a question. They get a confident, well-written, completely wrong answer, and then they act on it.

🥗 Nutrition advice is where it gets particularly unreliable

This is the area I feel most strongly about, because it sits right at the heart of what I do.

Nutrition is complicated. The research is constantly evolving, there is enormous individual variation in how people respond to food, and the gap between generic dietary advice and genuinely useful guidance is vast.

A study evaluating AI responses on dietary supplements found that overall accuracy was low across all modes tested, with nearly three quarters of citations coming from unverified sources such as blogs, sales websites, and social media. Around 10% of responses contained at least one citation where the source cited did not actually support the claim being made.

So when someone asks AI whether they should take a particular supplement, or what to eat to support their hormones, or whether a specific dietary approach is right for them, the answer they get may sound thorough and credible while being built on sources that no qualified practitioner would consider reliable.

Where AI falls particularly short is in addressing the human side of nutrition: our habits, emotions, and motivations, which are key to long-term success. And I would add to that: our individual biochemistry, our health history, our medications, our hormones, our stress levels, our gut health. All of the things that make nutrition advice actually useful rather than just generically correct.

🔍 AI doesn't know you, and that matters more than you might think

This is the crux of it, really. Generic AI is answering a generic version of your question. "Should I take iron supplements?" has a very different answer depending on your age, sex, and current iron status. Without that context, the answer won't be personalised or useful.

Take something like perimenopause. If you ask AI what to eat during perimenopause, you will get a list of foods. But for example, it won't know that you have a history of disordered eating, that you're also managing ADHD, that your last blood test showed your vitamin D was extremely low, or that you've tried a low calorie diet before and felt terrible on it. A good practitioner would know all of this, and it would change everything about the advice given.

The same goes for conditions like PCOS, PMS, PMDD, anxiety, gut problems, weight issues, emotional eating or anything involving mental health. These are areas where individual variation is enormous and where generic guidance can not only be unhelpful but can actively work against you.

One study also found that small differences in how symptoms are described can completely change the answer AI gives, making accuracy highly dependent on how well someone can articulate what they're experiencing. And when you're unwell, exhausted, or worried, precise articulation isn't always easy.

A note on privacy

This one often gets missed in the conversation. When you type a health question into a consumer AI chatbot, you are sharing personal health information with a platform that, in most cases, was not designed with healthcare privacy in mind. Symptoms, concerns about your mental health, questions about your cycle, worries about a diagnosis: these are some of the most personal things you will ever type anywhere, so it is worth being aware of where that information goes.

🍏 So what should you actually do?

Use AI as a starting point, not an end point. If it prompts you to ask better questions or helps you understand a term, great. 

But before you change something meaningful about your diet, start or stop a supplement, or make a decision about your health based on what an AI has told you, please sense-check it with a qualified professional who knows your full picture.

Because the difference between generic health information and advice that is actually tailored to you is enormous, and your health is worth that difference.

If you'd like to work through what's going on for you specifically, I'd love to help!

Click here to book a chat.