Recent research presented at the American Society for Nutrition's NUTRITION 2026 conference reveals that widely used AI-powered food apps may significantly underestimate calorie and fat content in meals.
Researchers evaluated four popular AI calorie-counting applications by analyzing 102 meals with precisely measured ingredients. The study found these apps underestimated calorie intake by an average of 250 to 345 calories per meal and fat content by about 30 grams.
Challenges in Accurate Estimation
The apps analyze meal photos by identifying foods, estimating portion sizes, and referencing nutrition databases. While this process is faster than manual entry, it struggles with accuracy, especially for meals high in hidden fats such as butter, oils, cheese, and sauces.
- Keto meals, rich in fats and low in carbs, showed the largest errors.
- Carbohydrate estimates were generally more accurate than calorie or fat counts.
- Meal size affected accuracy inconsistently across apps.
Implications for Users
Millions rely on these AI tools for weight loss, diabetes management, and healthy eating. Underestimations could hinder weight loss efforts and mislead users about their actual intake.
Experts advise that AI apps should be used as guides rather than precise measures. Individuals with medical conditions should continue following professional dietary advice and monitor portion sizes carefully.
Future Outlook
Ongoing research aims to improve AI recognition of hidden fats and complex ingredients. While AI advancements may enhance accuracy over time, current findings remain preliminary and await peer-reviewed publication.
For now, users are encouraged to view AI calorie counts as estimates and maintain a cautious approach to diet tracking.
