2025 was the last time we could see how algorithms work.

The model was prompted to deceive a person with an optical illusion. The resulting works reveal the inner mechanisms of a model incompetente to succeed due to lack of knowledge and experience. Such characteristic fails help us approach AI processes, but become fewer everyday with models being optimized for "correct" images.

Made with Stable Diffusion Flux.

3584x4608px | 21.4mb