The sameness problem behind those unappetizing AI-generated menus

AI-generated restaurant menus often exhibit a uniform, overly polished aesthetic because diffusion models and LLMs are trained on large, homogeneous corpora that favor “pleasing” visuals, such as legacy fast-food menus. This leads to convergence—repeatedly reproducing the same style—and risks model collapse when AI-generated outputs are fed back into training data, causing homogenized, sometimes uncanny food illustrations.