ChatGPT is not your nutritionist

Australian research published in 2024 found that around one in 10 people have asked ChatGPT a health question, a statistic that has likely grown as the popularity of generative AI continues to catapult. 

Worryingly, another study published in the Journal of Medical Internet Research found that ChatGPT advice was only consistent with expert, evidence-based recommendations 60% of the time. 

So, what are the risks of trusting your health – and your nutrition – to a tool that isn’t trained in clinical care or doesn’t understand the nuance of your situation? 

And what are some better ways of assessing reliable nutrition support? 

ChatGPT and nutrition advice: The risks 

To really understand the risks of using ChatGPT as an on-demand nutritionist, you need to consider how it works. 

ChatGPT trawls the internet for answers, pulling information from everywhere, including unverified sources and anecdotal evidence. This leaves room for inaccuracies. 

Think about it like this: Without doing your own research, there’s no way to tell if the information is coming from a scientific journal or a discussion on Reddit. 

Add to that, ChatGPT presents this information regardless of your individual needs or risks, so, for example, it could recommend the same ‘healthy’ breakfast advice for someone with insulin resistance versus someone without. 

That’s not the only area of concern… 

“It sounds smart, so it must be true”  

You might have noticed ChatGPT has a reputation for ‘hallucinating’ information. 

Many people claim that ChatGPT and other generative AI platforms have given them information from clinical studies and research papers that don’t actually exist. 

The platforms confidently deliver study findings and other pieces of information that, upon closer inspection, are purely fictional. 

These hallucinations were noted in a 2023 study that found ChatGPT provided an incorrect benefit of gallstone treatment. 

A lot of the information can be wrong, inaccurate, or out of date, too. The platform is known to have a knowledge ‘cut off’ date, so recent studies and findings won’t be included in its answers. 

Basically? It doesn’t have access to up-to-date clinical guidance like a nutritionist does. 

All of this underpins something important: that you can not take anything ChatGPT says at face value. Approaching it with a healthy scepticism and a lot of your own research is key. 

Credible sources, like scientific journals and peer-reviewed research, should take priority if you are conducting your own research about your health. 

And ultimately, it is health professionals, like nutritionists, who will give you the reliable, credible, and personalised knowledge you need to make informed choices about your health. 

The problem with prompts 

If you’ve played around with AI, you’ll know it’s only as good as what you tell it. The quality of ChatGPT’s responses will depend largely on the prompts you give it. 

If you don’t provide enough detail, or if the info you input is wrong, ChatGPT can’t fill in the gaps, so its advice will be incomplete or misleading.

This means, if you’re approaching it without an existing and in-depth understanding of health and nutrition, you’re going to get pretty lacklustre and potentially harmful results. 

Here’s an example… 

You might ask: “What supplements should I take for energy?” and ChatGPT will likely tell you to take B12 and iron. 

But it won’t it won’t ask whether you’ve had blood tests to check for deficiencies, whether you’re already taking other supplements or medications, or whether your low energy is actually linked to sleep, stress, thyroid issues, or an underlying condition.

This is why working with a nutritionist is key. They consider the individual and the bigger picture. 

The importance of nuance and connection 

You’re probably noticing a theme here. ChatGPT does not know your medical history. It doesn’t have access to clinical assessments or results. Nor does it understand your personal relationship with food. 

(This understanding is particularly important for anyone who has a history of disordered eating.) 

In fact, ChatGPT will regurgitate the same information to anyone who inputs a specific prompt, regardless of their health history, cultural food practices, or personal preferences. 

In short? It lacks nuance, which can be dangerous if you’re using it for queries related to nutrition and health. 

Let’s look at a couple of examples… 

Example 1 

Someone might ask: “What’s the best way to lose weight?”

ChatGPT could suggest a low-carb or keto diet, without knowing the person has a history of disordered eating, struggles with restrictive behaviours, or needs carbohydrates to support hormone health.

Example 2 

A person with diabetes or prediabetes might ask ChatGPT: “What foods should I eat to manage my blood sugar?”

ChatGPT might provide a list of low-glycemic foods or suggest reducing sugar intake without knowing their current medications, insulin regimen, or personal glucose response. 

ChatGPT is not your health hero 

ChatGPT may seem like a good place to go for nutrition advice. It’s quick and easy, and it seems to know its stuff, but AI nutrition advice isn’t tailored to your unique needs, risks, or medical history. 

And that lack of nuance can be risky for your health.

Bottom line? AI tools are super convenient, but they can not be relied on for diagnosis, treatment, or complex nutrition guidance. 

They’re no match for the personalised care, understanding, and expertise of a nutritionist.