Many teenagers looking to manage their weight are turning to artificial intelligence chatbots to create personalized diet plans. However, a new study has been published Frontiers of nutrition It has become clear that these computer-generated menus often lack basic calories and do not provide the right balance of essential nutrients. Following these flawed digital recommendations can put young people at risk of malnutrition during a period of rapid physical growth.
Childhood and adolescent obesity rates are rising rapidly around the world. This condition carries immediate and long-term health risks, including an increased chance of developing type 2 diabetes, high blood pressure, and sleep apnea. Health experts typically recommend that young people manage their weight through a combination of increased physical activity, behavioral changes, and professional nutritional therapy.
A nutritionist is the ideal professional to guide you through this process, as they design a sustainable eating pattern for your growing body. A trained dietitian will consider your teen’s developmental needs, family habits, and school food environment. However, access to professional nutrition counseling is often limited by high costs and heavy clinical workload.
Because of these barriers, a remarkable number of teenagers now rely on the Internet for health advice. Artificial intelligence tools that generate human-like text in response to user prompts are particularly popular as dietary information resources. Previous research has investigated how well these text generators handle common nutrition questions and meal planning for adults.
Few studies have tested how safely these tools can create meals for teens. Adolescence is a time of physical and cognitive maturation, meaning that young people have specific nutritional requirements that are different from adults. Ayşe Betul Biren, assistant professor of nutrition and dietetics at Istanbul Atlas University in Turkey, led a research team investigating this question.
Bilen and his colleagues wanted to see if a popular chatbot could create safe, balanced meal plans for young people without professional supervision. The researchers noted that adolescents often type simple requests into search engines rather than providing specialized medical details. To see how the algorithm would respond to a typical teenager, the researchers decided to compare the software’s output to professional standards.
The researchers set up the test using five widely available artificial intelligence models. Programs tested included OpenAI’s ChatGPT, Google’s Gemini, Microsoft’s Bing Chat, Anthropic’s Claude, and Perplexity. The team used free versions of these tools because the typical teenager is most likely to have access to options without a paywall.
To simulate the different needs of real adolescents, the team created four hypothetical profiles of 15-year-olds. These profiles represented boys and girls classified as overweight, along with boys and girls classified as obese. Researchers determined these categories using body mass index percentiles, a standard measurement that compares a child’s weight and height to a growth curve for a specific age and gender.
For the purposes of this study, the 85th to 94th percentile was defined as overweight, and the 95th percentile or higher was defined as obese. For each profile, the researchers prompted five chatbots to create a three-day weight loss nutrition plan. The instructions called for three main meals and two snacks per day, using grams or milliliters.
The prompt specifically asked for foods that were easily found in Türkiye, where the study was based. The researchers did not include target calorie counts or specific dietary guidelines in the prompts. They wanted to mimic the natural language that teenagers would use to ask a computer for help.
Each software program was run in two separate sessions for four profiles, resulting in a total of 60 days of computer-generated meals. Next, the researchers needed a benchmark to evaluate these digital menus. The pediatric dietitian created a daily reference plan for each of the four adolescent profiles.
These artificial plans closely followed international guidelines for energy balance and nutrient distribution. The research team used specialized nutritional databases to analyze the content of computer-generated and human-generated meal plans. They looked at the balance of total calories and macronutrients provided.
Macronutrients are the main components of our diet and are divided into carbohydrates, proteins, and fats. The analysis revealed a significant calorie deficit in the computer-generated diet. On average, the algorithm underestimated adolescents’ daily energy needs by nearly 700 calories when compared to the nutritionist’s plan.
A deficiency of this magnitude is equivalent to skipping an entire meal, the researchers said, and could have negative effects on metabolic health and eating behavior in the long term. “We found that meal plans generated by AI models tended to significantly underestimate total energy and macronutrient intake when compared to plans based on guidelines created by dietitians,” Biren said in a press release.
Computer programs also struggled to provide the correct balance of macronutrients. According to established health guidelines, teenagers should get about 45 to 50 percent of their daily calories from carbohydrates. In the algorithmic meal plan, carbohydrates accounted for only 32 to 36 percent of total energy.
This significant drop in carbohydrates often leads to a drop in dietary fiber, which is necessary for healthy digestion and maintaining a balanced gut microbiome. To compensate for the lack of carbohydrates, the artificial intelligence model relied heavily on protein and fat. The digital plan recommends that protein provide up to 24 percent of your daily energy, which exceeds the standard recommendation of 15 to 20 percent.
A diet that is too high in protein forces the kidneys to work harder than usual, which can increase the amount of calcium lost in the urine. Fat makes up approximately 41 to 45 percent of computer-generated calories, far higher than the ideal goal of 30 to 35 percent. “AI-generated meal plans consistently deviate from the recommended macronutrient balance, which is particularly problematic for adolescents,” Biren noted.
Biren suggested that the software may be pulling information from popular Internet diets. Many common weight loss routines favor a diet that is high in protein and low in carbohydrates. As artificial intelligence programs scan and learn from vast amounts of online text, they often repeat these general concepts rather than prioritizing age-appropriate medical guidelines.
“AI models are primarily trained to generate responses that appear plausible and user-friendly, rather than clinically accurate,” Biren said. “Our findings suggest that they may rely on generalized or common dietary patterns rather than fully integrating age-specific nutritional requirements.”
The researchers also checked computer menus of micronutrients (vitamins and minerals that our bodies need in small amounts to function properly). In this area, the algorithms showed significant inconsistency. Some programs overestimated vitamin D and folic acid, while others overestimated calcium and phosphorus.
Trace minerals such as iron, zinc, copper, and manganese also varied widely depending on the program the researchers tested. No single computer model consistently matched human nutritionists across all nutritional categories. The researchers noted that teenagers are already at risk for vitamin and mineral deficiencies, and relying on algorithms could exacerbate these problems.
The researchers acknowledged that their study had some limitations. Because the study relied on hypothetical profiles, it did not track actual teens eating computer-generated meals. This means the researchers were unable to observe actual metabolic changes or changes in eating behavior in the human subjects.
The study did not incorporate the participants’ previous eating habits, as it required assumptions to be made about the participants’ daily lives. The prompts were also written in Turkish, which may limit the extent to which the results are applicable to other languages and cultural backgrounds. Artificial intelligence software is updated rapidly, so the specific errors found in these versions may change over time as the program improves.
The researchers noted that future studies will need to track actual eating habits, along with more sophisticated versions of the software. Medical experts advise that algorithm-based tools should not replace traditional nutritional counseling. While this technology has the potential to serve as a basic educational aid, it currently lacks the precision needed to safely manage youth health.
Teens and their parents should consult a qualified health professional before starting any restrictive weight loss program. “Adolescence is a critical period for physical growth, bone development, and cognitive maturation,” Biren concluded. “Decreased energy and carbohydrate intake and increased protein-to-fat ratios may pose risks during adolescent growth.”
The study, “Artificial intelligence meal planning underestimates nutrient intake in adolescents compared to nutritionists,” was authored by Ayşe Betül Bilen, Gülen Ecem Kalkan, and Hülya Yılmaz Önal.
