An assortment of robot images generated by Stable Diffusion
Enlarge / An assortment of robot portraits generated by Stable Diffusion as found on the Lexica search engine.

According to Andy Baio of Waxy, some online art communities have taken dramatic steps to ban or reduce the amount of artificial intelligence generated artwork on their sites.

Baio noticed the bans on Friday and reported them on his website. Some vocal artists on social media have complained about how much art they see on those platforms, but major art communities have not made any changes to their policies.

The arrival of widely available image synthesis models such as Midjourney and Stable Diffusion has provoked an intense online battle between artists who view artificial intelligence as a form of theft and artists who embrace the new creative tools.

The artist communities are at a difficult crossroads because they fear non-ai artwork getting drowned out by an unlimited supply of artificial intelligence, and yet the tools have become popular among some of their members.

Newgrounds wrote that they wanted to keep the focus on art made by people and not have the Art Portal flooded with computer-generated art.

There are concerns about the ethics of how image synthesis models learn from existing artwork. We don't think it's in our community's best interests to allow artificial intelligence on the site.

The debate about how art communities can adapt to software that can potentially produce unlimited works of beautiful art at a rate that no human working without the tools could match is only the most recent move.

Advertisement

Part of a wider art ethics debate

An array of non-AI artworks used to train Stable Diffusion, assembled by Waxy.org.
Enlarge / An array of non-AI artworks used to train Stable Diffusion, assembled by Waxy.org.

The current wave of image synthesis tools allow users to type in a written description and then output a matching image. With a skillfully crafted prompt, the results can mimic the works of human artists with sometimes stunning detail.

Most of the time, the prompt references existing artists and art websites. New styles can be created by mixing artists. The robotic woman is in the center of the image at the top of the article.

Beautiful crying! female mechanical android!, half portrait, intricate detailed environment, photorealistic!, intricate, elegant, highly detailed, digital painting, artstation, concept art, smooth, sharp focus, illustration, art by artgerm and greg rutkowski and alphonse mucha (Seed 79409656)

The most popular image synthesis models are able to analyze billions of images. The images come from the LAION-5B database, which is located on the internet. It's ideal for training artificial intelligence models to have descriptions attached to images found on the internet.

Baio and Simon Willison created a search tool that allows users to look into a small slice of the larger set after grabbing data for more than 12 million images. You can use the LAION5B image set to search for artwork or even your own name.

A search for Jeremy Lipking, a living fine art painter, brings up results in the LAION data set.
Enlarge / A search for Jeremy Lipking, a living fine art painter, brings up results in the LAION data set.

Some artists were unhappy about their artwork being discovered in the Stable Diffusion data set. A detailed report about these reactions was written by Charlie Warzel. With battle lines drawn firmly in the sand and new artificial intelligence tools coming out steadily, this debate will likely continue for some time to come.