Computer art: is AI-generated art stealing human artists’ work…
We are still in uncharted territory
Artificial Intelligence (AI) programs that generate art can be trained by learning from a database of human work. This can be made up of images in the public domain. But there are many contemporary artists that have found themselves in AI databases without copyright permission- and they can’t easily opt out or get paid for their work.
This has prompted Berlin artist pioneers Holly Herndon and Mat Dryhurst to create a website for artists to find out if their work has been included in computer learning datasets called Have I Been Trained.
The idea is that an artist can search generative AI datasets for their own work and then ask to be removed from LAION, a non-profit that makes enormous machine learning models. That way, artists can choose whether their work can be used to train AI.
When I think about it- I think about how I started to learn to draw. One of the many ways to build skill is to copy old masters. Here’s my first attempt to make a copy of an old master:
Not too bad- but my art prof told me what’s wrong with it in real time- and let’s just say, it wasn’t in danger of passing as a forgery!
This is one way for people to learn how to draw: copy the old masters.
So why shouldn’t a machine use an old master to learn how to draw?
For that matter, why can’t a machine use any art it is programmed to analyze in order to learn how to draw and then take instruction from its human master- kind of like contemporary artists that outsource to assistants?
As long as the AI is generating something original, even if it is based on skills acquired from copying perhaps millions of human-made works, how is that different than how anyone has every learned anything?
As with most artistic endeavors, it’s difficult verging on impossible to monetize fine artwork. So we’ve got a conundrum- if the art is being made and no one else cares, or cares to pay for it, what should be the cost of copying it. And how much is reasonably spent enforcing an artist’s copyright.
When you have something like French art group “Obvious” that sells an AI generated piece of art at Christie’s for $430,000.00, people start to take notice.
But the U.S. Copyright Office isn’t so keen to copyright art made by AI.
When Dr. Stephen Thaler filed an application at the U.S. Copyright Office to register a work of art called “A Recent Entrance to Paradise,” he was denied. Why? Because the author wasn’t a human. Dr. Thaler is continuing his quest to register a copyright through the courts.
As with almost anything that enters commerce, it can (and will) be assigned a value by the market. If there was a way to take one of the million artworks in a data set that a computer was trained on which then created a ten million dollar piece of art, should that artist receive ten dollars when that work is sold? What is the cost of inspiration? Does it matter if that inspiration was chosen for the AI by a programmer or at random? What about the creation and artistry of the programmer who directed the AI? From learning to product, there is a long road to haul, with its own costs and creative input.
I’m interested to see how computer-generated art continues to evolve, and how our talented human programmers and non-computer artists respond.
This article is for informational purposes only. It should not be considered Financial or Legal Advice. As much as I may try, not all information may be accurate.
