AI is creating bionic artists
Match-ups between humans and computers were a regular feature of the Cold War chess-tournament circuit. Computers were consistently beaten by human players for half a century, until, one day in 1997, humanity lost. Further research into decision-making would establish a surprising trend. If a human is working in concert with a computer, the pair performs better than the individuals would separately.
This is an important lesson that one should keep in mind whenever a new technology surfaces, including artificial intelligence (AI). The filmmaking industry has already found ways to use AI, up to the point where entire movies can be made using it — but only if creative humans guide it, from script writing through final output. Currently, Mechanism Digital is using an AI in the early stages of a project, mainly to produce inspirational imagery. In another project, we'll be using AI to produce matte paintings for background imagery.
Greg Rutkoswki is an artist whose opinion of AI generators has grown increasingly negative over time. Rutkowski is one of the few living artists that can attest to what it feels like to be source material in an AI’s brain. When an AI famously produced a work that strongly echoed Rutkowski’s Revolution, people noticed.
It’s possible to review the manually input prompts given to Stable AI, and it was revealed that 100,000 humans sat in front of a keyboard and typed Rutkowski’s name — compared to about 3,000 who were seeking images in the style of Pablo Picasso. This drives home the point that people are the ones pulling the strings of AI generators.
Behind the scenes, Stable Diffusion’s parent company is engineering its application to work well within other applications. An API, or Application Programming Interface, is being developed by Stability AI with plans on licensing it to other companies.
When I wrote "Prompting AI Imagery for Production," I raised the question of whether the three big AI generators — Stable Diffusion, DALL-E, or Google Dream, are capable of making art. At first, I didn’t want to call AI-generated images “art,” because the computer wasn’t trying to evoke an emotion. Then I realized the computer is just the tool — and the artist is the person who writes the prompt that is rendered by the computer. Prompt crafting is not so simple, and they usually have to be rewritten many times until the computer produces an image the user is happy with. Today I think AI-generated Imagery is art. It doesn’t require the talent of a fine art painter, but more like the talent of a photographer who finds, or sets up, a meaningful or beautiful picture before capturing the image using technology.
Originally published: 01/25/2023