AI Nutrition Advice: Can AI Be Trusted?
A Registered Dietitian’s Perspective
Written by: Riley Hughes, PhD, RD, CF-L1 | Dietitian and Fitness Coach at Experience Momentum
Curious about using AI for nutrition advice? Us too!
TLDR: At least at the moment, AI still falls short in its ability to provide comprehensive, appropriate, and accurate nutrition advice for individuals. These tools can fabricate information and references, posing a risk for the spread of misinformation and potential harm to individuals following its recommendations.
AI also struggles to account for individual clinical complexity, cultural food preferences, as well as emotional and motivational aspects of dietary counseling. That doesn’t mean AI tools have no place or use in your nutrition journey though! AI can be a useful tool and starting point for your nutrition questions (and finding recipe ideas), but for now we recommend booking an appointment with one of our Registered Dietitians to double check any information and advice you get from AI.
Now for the long version…
It seems like every other post on social media or news article seems to be about what to eat or not to eat, a new supplement to take, or explaining how cortisol is the root of every nagging symptom you’ve been googling recently (or is this just my social media feed?). Regardless, it is overwhelming to try to decipher what nutrition advice is accurate, misleading, or just outright false. And when AI is just a click and type away, it makes sense to see if it can clear up some of your confusion about what or how much to eat.
What the research says
AI can process information, but (human) experts determine meaning, clinical relevance, and application of information. The highest risk of misinformation comes from asking AI to generate information, rather than interpret or summarize it. When answers to medical questions were assessed for accuracy, nutrition ranked among the lowest. Overall, accuracy was <50% for dietary advice and got worse when used in more complex cases that involve multiple medical conditions. However, all answers were expressed with confidence and certainty by the chatbots. To support its claims, AI may also fabricate non-existent references (“hallucinations”). Studies show that up to 60% of citations from some of these tools may be fake. Therefore, it is not only crucial to verify that the sources say what AI claims they do, but also that they exist in the first place!
Additionally, when meal plans made by AI tools were compared to plans made by registered dietitians, AI was found to underestimate energy needs and did not provide appropriate recommendations for adequate macro or micronutrient intake. This was found in particular for adolescents, which is alarming and could lead to detrimental impacts on growth and development.
Limitations and risks
Use of fake citations leads to the propagation of misinformation and may result in unsafe dietary advice or misinterpretation of health conditions. Would you trust a doctor that you knew was making up 60% of the information they were giving you? Probably not…
AI also does not assess source validity or study quality. While some specialized AI tools are trained on a specific dataset, most widely available AI tools draw on all information and do not hierarchize sources based on quality. Rather, information appearing more frequently or in more contexts receives greater implicit weight, not because AI judges it as more reliable, but because it is statistically more reinforced. This means that the hundreds of Reddit posts you saw while doomscrolling in bed last night get ranked alongside the peer reviewed article published in Nature. This can lead to the promotion of popular, fad diet trends without assessment of safety or validity. AI models that respond to human feedback are also known to exhibit sycophancy, providing answers that align with user beliefs over the truth.
Underestimation of energy needs and lack of adequate consideration to micro and macronutrient balance can lead to fatigue at best and health risks of nutrient deficiencies or development of disordered eating at worst. It is incredibly important to discuss your nutrition needs with a dietitian to ensure you are getting enough to maintain energy, strength and endurance for your physical activities, hormone regulation, and lots of other important body processes!
Potential uses and benefits
Struggling to come up with a dinner idea with the items in your fridge? AI can help with that. Let it know your ingredients and how much time you have and it can come up with some recipe ideas for you!
Struggling to understand the jargon of a scientific article you found or want to get a summary of multiple articles? AI can help with that. Summarizing and compiling information is where AI excels and can be of primary benefit for users.
So if you do decide to play around with AI and ask it some questions about nutrition, do so with plenty of grains of salt and please talk with your dietitian about anything you find. We’re as curious as you and we want to know what information you’re getting so we can help assess it and provide clarification or nuance to any advice or claims that it is making.
References
Bilen, A. B., Kalkan, G. E., & Önal, H. Y. (2026). Artificial intelligence diet plans underestimate nutrient intake compared to dietitians in adolescents. Frontiers in nutrition, 13, 1765598. https://doi.org/10.3389/fnut.2026.1765598
Ponzo, V., Rosato, R., Scigliano, M. C., Onida, M., Cossai, S., De Vecchi, M., Devecchi, A., Goitre, I., Favaro, E., Merlo, F. D., Sergi, D., & Bo, S. (2024). Comparison of the Accuracy, Completeness, Reproducibility, and Consistency of Different AI Chatbots in Providing Nutritional Advice: An Exploratory Study. Journal of clinical medicine, 13(24), 7810. https://doi.org/10.3390/jcm13247810
Shah, N. H., Entwistle, D., & Pfeffer, M. A. (2023). Creation and Adoption of Large Language Models in Medicine. JAMA, 330(9), 866–869. https://doi.org/10.1001/jama.2023.14217
Tiller, N. B., Marcon, A. R., Zenone, M., Kidd, K. E., Jeukendrup, A. E., Master, Z., & Caulfield, T. (2026). Generative artificial intelligence-driven chatbots and medical misinformation: an accuracy, referencing and readability audit. BMJ open, 16(4), e112695. https://doi.org/10.1136/bmjopen-2025-112695
Walters, W. H., & Wilder, E. I. (2023). Fabrication and errors in the bibliographic citations generated by ChatGPT. Scientific reports, 13(1), 14045. https://doi.org/10.1038/s41598-023-41032-5