New research suggests that popular AI tools may be giving incomplete and unbalanced dietary advice to teens, raising important questions about whether these technologies are ready to guide growing bodies without expert oversight.
Study: Artificial intelligence diet plans underestimate nutrient intake in adolescents compared to dietitians. Image credit: ilona.shorokhova/Shutterstock.com
Artificial intelligence (AI) is increasingly being used for diet planning among teens, but a new study suggests it may fall short of expectations. investigation, published in leader in nutritionfound that AI-assisted recommendations may consistently underestimate the required nutritional intake for adolescents.
Rising obesity among adolescents increases demand for accessible dietary advice
Overweight and obesity rates among adolescents are increasing rapidly globally, affecting approximately 390 million adolescents in 2022. In fact, many regions now report it as the leading form of malnutrition. Excessive body weight is associated with several adverse health outcomes, including type 2 diabetes, abnormalities in blood cholesterol, high blood pressure, and sleep apnea. These youth are more likely to be obese as adults and have a lower quality of life.
Teens also have body image concerns and a desire to lose weight, which includes potentially dangerous methods such as vomiting after meals or excessive use of laxatives.
Dietary modification is important to improve the health of children in this area. Dietitians are health professionals who design and supervise individual nutrition plans according to established guidelines. However, their services are not always accessible, and their heavy workloads may prevent teens from receiving needed dietary advice and follow-up.
AI-based tools such as chatbots are being used to overcome these limitations, but only a few studies have evaluated their role in adolescent nutrition. Similarly, large language models (LLMs) such as ChatGPT can provide useful information to support nutritional planning, but with important limitations.
Existing research indicates that they may not meet safety standards or international nutritional recommendations, especially in real-world conditions. AI tools are also unlikely to provide the same level of patient services that dietitians do. However, most of this evidence is based on adult studies or clinical cases.
The current study sought to directly compare an AI-generated diet with a diet designed by a personal dietitian for overweight or obese adolescents. The areas of comparison were energy and nutrient content, safety and feasibility. The comparison could show whether AI chatbots could replace dietitians in nutrition planning for this patient category or be used as adjuncts under dietitian supervision.
Researchers compare five AI tools with dietitian plans
The researchers used five AI models (ChatGPT-4O, Gemini 2.5 Pro, Cloud 4.1, Bing Chat-5GPT, and Perplexity) to generate 60 diet plans over two sessions. Three-day diet plans were created by each model in response to prompts using four standardized adolescent profiles: a boy overweight or obese, and a girl overweight or obese.
These were compared to a reference one-day diet plan prepared by a dietitian for each profile. It followed nutritional recommendations with energy distribution as follows: 45–50% from carbohydrates, 30–35% from lipids and 15–20% from proteins.
The researchers then analyzed the energy and macronutrient content of each plan.
AI underestimates dietary energy and key nutrients
The results revealed a consistent and potentially worrying pattern. The AI model included less energy and macronutrients than dietitians. The energy deficit was 695 kcal, while protein was 20 g less, fat was 16 g less and carbohydrates were 115 g less. The potential energy difference may have important clinical implications, especially given the high energy demands of adolescents.
The authors suggest that, given this general oversupply of fat and low carbohydrate content, LLMs may rely more on popular diets such as the ketogenic diet than on scientific guidelines, which explain the low-carbohydrate, high-fat approach. This can impair growth, metabolism, and cognitive development during this critical developmental window. Thus the long-term safety of such recommendations is unproven.
Five models recommended protein content up to 23.7% and fat content up to 44.5%. Both were above recommended levels for teenagers. In contrast, carbohydrate content in the diet was a maximum of 36.3%, which was below the recommended level.
Dietitian plans included 44%–46% carbohydrates, depending on the profile. The protein percentage varies between 18% and 20% and the fat between 36% and 37%. Overall, these plans were in line with national recommendations.
The authors explain that “This pattern reflects a systematic shift toward lower CHO, higher protein, and higher lipid food compositions across all AI models, indicating that macronutrient balance, not just gram-based nutrient intake, is substantially disrupted in AI-generated plans.”
The micronutrient composition of AI-generated diets varies considerably, With notable variability between models and compared to the dietitian reference. This may contribute to micronutrient deficiencies in adolescents, indicating that these plans may not be suitable for clinical use without professional supervision. Neither model closely followed the dietitian reference diet in all nutrients.
The authors say this is the first time that different LLMs have been compared for nutritional needs in adolescents, including a detailed assessment of macro- and multiple micronutrients as well as macro-nutrient needs. As earlier research suggests, this may indicate a lack of AI technical expertise in this area. This may hinder accurate estimation of energy and macronutrient composition in AI-generated personalized diet plans.
Strengths and limitations
The study has several strengths. It evaluated five different AI models, increasing the robustness and comparative power of the analysis. By creating three-day diet plans, the researchers were able to assess consistent patterns rather than isolated anomalies, strengthening the reliability of the findings. The use of dietitian-designed plans based on international dietary guidelines provided a reliable and clinically relevant reference standard. Furthermore, comprehensive assessment of macro- and micronutrients enabled a detailed, multidimensional assessment of diet quality.
Despite these strengths, the study also has limitations. Its findings may only apply to the specific AI models tested, which are constantly evolving, and some potentially relevant information may be absent from standardized teen profiles, limiting personalization. Statistical approach, including the use of averaged multi-day outputs, may affect the independence and variability estimates of the results. Furthermore, the study relied on simulated scenarios rather than real-world adolescent behavior, which may limit ecological validity. Finally, the use of standardized prompts in a single language may restrict the generalizability of the findings to other populations and settings.
Risks of untrained AI nutrition advice
“AI models have demonstrated clinically significant divergence in dietary plans for adolescents at both the macro and micro levels.” He consistently recommended a diet with lower energy and carbohydrate content than dietitian-designed diets.
Until these shortcomings are addressed, the authors caution that AI-generated diet plans should not replace professional dietary guidance for adolescents.
