When you first encounter an AI-generated menu, it can feel unsettling. You might find yourself in a cafe, staring at a menu filled with bagel sandwiches, each illustration appearing eerily flawless and oddly smooth. This phenomenon is not a figment of your imagination; generative AI menus are making their way into the restaurant industry, influenced by models trained on a narrow aesthetic that often feels off.
The Aesthetic Dilemma
Alex Lisle, CTO of Reality Defender, describes the situation: "It’s almost like an alien trying to make a pizza without understanding its core principles." Reality Defender is part of a growing sector of startups focused on AI detection and content verification, a response to the issues arising from AI-generated content.
The illustrations often appear too perfect, with every ice cream scoop looking flawlessly round and shrimp seemingly modified to consume their own tails, resulting in what Lisle calls "Lovecraftian food horrors."
How AI Models Work
Large language models (LLMs) and diffusion models, which power chatbots and image generators like ChatGPT and Midjourney, are trained on vast datasets. These models identify patterns to predict user requests, such as generating a menu for a burger restaurant. Lisle notes, "A lot of this stuff looks like a Chili’s menu from 2015, and there’s a reason for that."
