While AI-powered food tracking apps have simplified the process of logging meals, new testing suggests these tools may be significantly less accurate than users assume. A recent evaluation of four popular applications found that they underestimated calorie and fat content by approximately one-third when compared to precisely prepared laboratory meals.
The Challenge of Ketogenic and High-Fat Foods
The study highlighted a specific struggle with high-fat, ketogenic dishes. While the AI models were relatively consistent when measuring carbohydrates, they frequently failed to account for the dense caloric profile of fats. In some instances, the discrepancy reached as high as 345 calories per meal.
This margin of error presents a challenge for individuals relying on these apps for medical or weight-management goals. As the technology evolves, experts suggest using these tools as general guides rather than absolute measurements of nutritional intake.
