Comment by joebo

It seems the problem is looking for a single picture to represent the whole. Why not have generative AI always generate multiple images (or a collage) that are forced to be different? Only after that collage has been generated can the user choose to generate a single image.