Sam Biddle thought he’d stumbled on a badly cooked shrimp scampi. Zooming into the photo posted outside a New York deli he visited in August, the tech reporter for The Intercept instead recognized something closer to tentacled creatures out of an H.P. Lovecraft story. The dish never existed: the photo, like a growing number of images on American menus, had been generated by artificial intelligence.
The trend has a name on social media: food “slop,” after the mash fed to pigs. TikTok accounts have spent months collecting these images, burgers with impossible proportions, sushi in fluorescent colors, pasta with textures that defy description, often posted by restaurants, food trucks and delis that don’t have the time or the budget for a real photo shoot. According to the National Restaurant Association, 42% of US restaurant operators failed to turn a profit in 2025, and 26% of industry professionals already use AI, mostly for marketing.
That leaves one question: why do these particular images turn stomachs so effectively? Researchers point to the “uncanny valley,” the unease first described in 1970 by Japanese roboticist Masahiro Mori around robots that look almost, but not quite, human. A 2025 study led by psychologist Alexander Diel found that slightly imperfect AI-generated food images were rated as more disturbing and less appetizing than either clearly unrealistic images or fully realistic photographs, even though people judged them to have the same nutritional qualities on paper.
Two theories try to explain the reaction: an old instinct against food that might be spoiled, or food neophobia, the wariness that kicks in around anything unfamiliar. Either way, the lesson for restaurant owners tempted by the shortcut is simple: on a menu, the human eye spots a near-miss far faster than it reads a daily special.