Generative AI got left alone in a room with itself — text-to-image, image-to-text, rinse, repeat — and what started as dramatic prompts like stressed prime ministers plotting peace deals devolved into endless loops of bland city skylines, fancy empty rooms, and pastoral yawn-fests researchers dubbed “visual elevator music.” No drama, no people, no edge — just polished, soulless generics that forget their origins faster than you forget last week’s Netflix binge. The study proves it: without humans slapping the wheel, AI doesn’t innovate; it homogenizes into the safest, most average slop imaginable, raising the spooky flag that our culture’s already sliding into a loop of pretty-but-pointless mediocrity.