AI Food Apps Miss One-Third of Calories, Study Finds

New research presented at the American Society for Nutrition indicates that AI-powered calorie-tracking applications frequently fail to accurately estimate food intake ^[1]. The study found these tools often underestimate caloric and fat content by approximately one-third ^[1]. Findings showed that some applications were off by as much as 345 calories per single meal ^[1].

Ketogenic dishes posed the greatest challenge for the AI algorithms tested in the study ^[1]. While high-fat meals are popular among many dieters, the software often struggled to recognize the energy density of these specific ingredients ^[1]. Experts suggest users should remain cautious when relying solely on automated tracking for weight management or health goals ^[1].