Artificial intelligence tools used to generate diet plans for adolescents may not provide adequate nutrition, according to research published in Frontiers in Nutrition on Mar. 20. The study compared AI-generated meal plans with those created by professional dietitians and found that the technology often underestimates the nutritional requirements of teenagers.
This issue is important as adolescent overweight and obesity rates are rising globally, affecting about 390 million young people in 2022. Proper dietary guidance is crucial during this developmental stage, but access to qualified dietitians can be limited, leading some to turn to AI-based solutions.
Researchers evaluated five popular AI models—ChatGPT-4o, Gemini 2.5 Pro, Claude 4.1, Bing Chat-5GPT, and Perplexity—by having them create three-day meal plans for four standardized adolescent profiles. These were then compared with reference diets prepared by a registered dietitian following established nutritional guidelines.
The findings showed that all tested AI models produced diets lower in total energy and macronutrients than those recommended by the dietitian. Specifically, there was an average shortfall of 695 kilocalories per day, along with reduced amounts of protein (20 grams less), fat (16 grams less), and carbohydrates (115 grams less). The authors said this could have significant clinical implications given adolescents' high energy demands.
The study also found that AI-generated diets tended to favor higher fat and protein content while reducing carbohydrates below recommended levels. "This pattern illustrates a systematic shift across all AI models to lower CHO, higher protein, and higher lipid meal structures, indicating that the macronutrient balance, not just the amount of gram-based nutrients, is significantly disrupted in AI-generated plans," the authors said.
Micronutrient content varied widely among the different AI-generated diets and did not consistently match the reference plan from a dietitian. The researchers noted this variability could lead to micronutrient deficiencies if such plans were followed without professional oversight.
While the study's strengths include its use of multiple AI models and comprehensive nutrient analysis over several days, limitations exist. These include reliance on simulated scenarios rather than real-world behaviors and potential lack of personalization due to standardized profiles.
"AI models have exhibited clinically significant deviations in diet plans for adolescents at both macro and micro levels." Until these issues are resolved, the authors caution against using AI-generated meal plans as a substitute for professional dietary advice.
