Rutkowski was surprised at first, but decided it might be a good way to reach a new audience. He then tried searching for his name to see if a piece he was working on had been published. An online search returned a job with his name attached, but it wasn’t his.
“It’s only been a month. What about in a year? I probably won’t be able to get my job there because [the internet] will be flooded with artificial intelligence,” says Rutkowski. “That’s worrying.”
Stability.AI, the company behind Stable Diffusion, trained the model on the LAION-5B dataset, which was compiled by the German nonprofit LAION. LAION collected the data set and narrowed it down by filtering out watermarked images and non-aesthetic ones, such as logo images, says Andy Baio, a technologist and writer who downloaded and analyzed some of Stable Diffusion’s data. Baio analyzed 12 million of the 600 million images used to train the model and found that many of them came from third-party websites like Pinterest and art shopping sites like Fine Art America.
Many of Rutkowski’s works are drawn from ArtStation, a website where many artists upload their online portfolios. Its popularity as an AI prompt stems from a number of reasons.
First, its fantastical and ethereal style looks very cool. He’s also prolific, and many of his illustrations are available online in high enough quality that there are plenty of examples to choose from. An early text-to-image generator called Disco Diffusion offered Rutkowski as an example prompt.
Rutkowski has also added alt text in English when he uploads his work online. These image descriptions are useful for visually impaired people using screen reader software and help search engines rank images. This also makes them easy to scrape and the AI model knows which images are appropriate for the prompts.