While this story may not be remarkable, and it seems mostly forgotten as it has only been reprinted once since 1991, it does offer a quaint perspective on current “complicated” relationship between generative AI models and art.

Analog January 1991

Nina is an artist and works with paintings. As all artists she is struggling to make a name for herself, experimenting with a lot of abstract surreal paintings. Her apartment is rather messy, so her friend Eugene gives her some of his experimental nanomachines as they can actually help with the cleaning. They work by having some default patterns which they aim to restore, so they are able to discern that breadcrumbs are not supposed to be on the floor and so forth.

The nanomachines are working as intended, but unsurprisingly they create problems with her artworks as the machines also restores the paintings to a “default pattern” - basically making them realistic again. Of course, what may sound like a setback for Nina, by random chance she manages to turn it around to her advantage in the end.

Current generative AI models are more than capable on making weird surreal images. In fact, they struggle more with creating something realistic, but they general concept of using machine learned patterns to create something was an amusing concept from a story that is more than 30 years old.

Read in Analog January 1991
Rating: 3+