Ecological and nutritional analysis of artificial intelligence-based sustainable dietary models

Köktürk S. N., Yardımcı H.

JOURNAL OF FOOD COMPOSITION AND ANALYSIS, cilt.149, sa.2026, ss.1-12, 2026 (SCI-Expanded)

Özet

Growing concerns over environmental sustainability and health have increased the demand for sustainable diets.

Advances in artificial intelligence (AI) offer new opportunities for diet planning. Evaluating the nutritional

adequacy and environmental impact of AI-generated sustainable diets is essential. This study aimed to compare

the nutritional quality and ecological footprints of sustainable diets created by four popular AI tools (Gemini,

Copilot, ChatGPT, and Grok). Each AI tool generated a sustainable diet plan for a healthy adult woman. Diets

were evaluated for energy, nutrient content, antioxidant capacity, glycemic index, protein quality, fatty acid

profiles, Healthy Eating Index-2015 scores, and ecological footprints. All diets showed low ecological footprints

but did not meet energy targets (1800 kcal/day). Gemini had the most sustainable composition. Vitamin B12 was

consistently below recommended levels in diets from Gemini, Copilot, and ChatGPT. All diets had adequate

carbohydrate, protein, fat quality, and overall diet quality. Carbon footprint associated with protein, energy from

protein, cholesterol, pyridoxine, folate, and vitamin C. Water footprint was linked to multiple nutrients, and

fruits, meat, poultry, and fish; vegetable intake correlated with carbon footprint. AI-assisted diets have potential

for sustainability but nutritional completeness concerns remain, especially for vulnerable groups. Further studies

comparing AI and dietitian plans are needed to assess long-term effects.