Can ChatGPT generate energy, macro- and micro-nutrient sufficient meal plans for different dietary patterns?

Abstract This study evaluated the nutritional adequacy of dietary meal plans generated by ChatGPT, an AI program for natural language processing. Meal plans were generated for three different dietary patterns, omnivorous, vegetarian and vegan, and their nutrient content was analyzed and compared to reference values. All meal plans exceeded the recommendations for protein and most failed to meet the recommendations for carbohydrates and fat. Saturated fat and fiber were within recommendations as were most micronutrients, except for vitamin D, fluoride, iodine, and selenium. The vegan meal plan lacked vitamin B12, and ChatGPT did not suggest supplementation. Overall, our findings indicate that ChatGPT cannot generate nutritionally adequate meal plans for the three dietary patterns evaluated in this study. It can, however, be a useful tool to generate meal plans that are then evaluated and adjusted according to the reference values.