The Sameness Problem of AI Images in Restaurant Menus

Serdar HocamAuthor & Editor

AI-generated menu images starting to be used in restaurants harbor an unsettling homogeneity due to narrow datasets and repeated edits.

◉ 0 views
The sameness problem behind those unappetizing AI-generated menus | TechCrunch

AI-powered food images, which are becoming widespread in cafes and restaurants, create a uniformity problem that evokes a sense of unease in consumers with their overly flawless and symmetrical structures.

Perfection in AI Menus

While bagel sandwich illustrations on restaurant menus draw attention with their flawless symmetry and strange smoothness, they create an unusual feeling in humans.

This situation leads models, trained according to a narrow and pleasing aesthetic understanding, to resemble a stranger trying to produce food without grasping the basic principles.

Model Collapse and Data Loop

The over-feeding of AI models through content they produce themselves brings along the risk of model collapse or convergence, which lowers output quality.

The repetition of local adjustments, such as restaurants constantly changing prices or product names, also causes food images to become even more round and smooth over time.

Uncanny Valley Effect

Researchers from the University of Duisburg-Essen in Germany have determined that AI-made food images create an uncanny valley effect.

These overly realistic artificial visuals evoke significantly more displeasure and disgust in humans compared to completely fake images.