Artificial intelligence tools are increasingly being used to generate meal plans and diet advice. But a new study suggests that when it comes to teenagers trying to lose weight, those plans may not always provide the nutrition growing bodies need.
Researchers analyzing several popular AI models found that meal plans generated for adolescents consistently underestimated calorie needs and produced imbalanced nutrient profiles when compared with plans designed by a registered dietitian.
The findings were published in Frontiers in Nutrition.
“We show that diet plans generated by AI models tend to substantially underestimate total energy and key nutrient intake when compared to guideline-based plans prepared by a dietitian,” said Dr. Ayşe Betül Bilen, an assistant professor at Istanbul Atlas University’s Faculty of Health Sciences.
The research team asked five widely used AI tools to create meal plans for teenagers attempting to lose weight. The prompts included information such as age, height and weight and requested three days of meals with three meals and two snacks per day.
The resulting plans were created for four hypothetical 15-year-old teenagers, including both boys and girls in the overweight and obese percentiles.
Researchers then compared the AI-generated plans with meal plans prepared by a dietitian specializing in adolescent health.
On average, the AI-generated plans underestimated daily energy needs by nearly 700 calories — roughly the equivalent of skipping a meal.
That difference could have meaningful consequences for adolescents, whose bodies are still developing.
“Following such unbalanced or overly restrictive meal plans during the teenage years may negatively affect growth, metabolic health, and eating behaviors,” Bilen said.
Beyond calories, the study also found that AI-generated plans produced macronutrient ratios that differed from standard nutrition guidelines.
The plans typically included higher levels of protein and fat while providing substantially fewer carbohydrates than recommended for adolescents.
Protein intake in the AI-generated plans was about 20 grams higher than in the dietitian-designed plans. Fat intake was also elevated, accounting for roughly 41–45% of calories.
Carbohydrates, however, were much lower than recommended levels, with the AI-generated plans averaging about 115 grams less carbohydrate than the dietitian’s plans.
For comparison, nutrition guidelines from the National Academies of Sciences, Engineering, and Medicine suggest that carbohydrates should provide about 45–50% of total energy intake, with 15–20% from protein and 30–35% from fats.
The researchers suggest that the discrepancy may stem from how AI models are trained.
“AI models are primarily trained to generate responses that appear plausible and user-friendly rather than clinically precise,” Bilen said. “Our findings suggest they may rely on generalized or popular diet patterns instead of fully integrating age-specific nutritional requirements.”
The researchers note that not all teenagers have access to professional dietary guidance, which may lead some to turn to AI tools for meal planning.
But they caution that overly restrictive diets or extreme macronutrient patterns could be particularly problematic during adolescence, a period of rapid physical growth and brain development.
“Adolescence is a critical period for physical growth, bone development, and cognitive maturation,” Bilen said. “Lower energy and carbohydrate intake, combined with increased protein and fat ratios, may pose risks during the adolescent growth period.”
The authors say AI tools may still be useful as educational resources, but they should complement — not replace — professional dietary guidance, especially for vulnerable groups such as teenagers.
As AI technologies continue to evolve, the researchers say improving their alignment with evidence-based nutrition guidelines could help make these tools safer and more reliable for people seeking diet advice.
The researchers reported no commercial or financial conflicts of interest related to the study.