From my perspective AI is a tool that artists can use, rather than a replacement. The main pitfalls are if it were to use an artist’s work without compensation, although there seems to be solutions.
If a program uses the work of a copyrighted artist without compensation, that would be wrong. But there should be ways to have AI-generated art that doesn’t come with that problem. Hiring young artists to churn out examples for the model (and taking advantage of public domain material) seems to be the easiest fix to the “theft” argument.
It still comes with the problem that it’s a new form of competition for professional artists, but I don’t support the argument for stagnation.
Comments about how artists may suffer seem to be like the libertarian complaint that if the lightbulb were invented today, the candlemakers would call for legislation protecting them. Or it would be like painters complaining about the invention of the camera.
Obviously if there’s theft involved, this adds insult to injury. But if you can do it without theft, there are a lot of possibilities. Some people will lose as with any change. They have specialized skills that might not translate to an environment where new tools are available.
But talented people will be able to produce more material in less time. A comics artist may be able to draw 50 pages instead of 20. A vfx team can do more. This will allow more types of stories to be told, if special effects budgets can be cheaper. Or it could also be that we’ll have a race for spectacle, and the budgets will be the same, but the effects will be bigger.
There’s a feat that it will put a lot of people out of work, I don’t think a comic book artist drawing 60 pages a month instead of 20 would put two other artists out of work. If an artist can work faster, they can appeal to a smaller audience or tell stories differently (scenes can be allowed to breathe if the cost of additional pages is reduced.) 1-2 people making the work of hundreds are just not going to produce good enough work.
I don’t think AI will get good enough that a handful of people can make a 3D animated film. You’re going to need humans to sand the rough edges especially in film and comics, where it’s not just about having one image.
If AI gets good enough that someone with a program can make a movie that would’ve normally employed a hundred people, there’s not much that can be done to stop it. Pressuring Disney won’t stop independent outfits, and changes to American laws won’t stop China (though I suppose China might be nervous about whether they can control AI.)
I wonder if a subtext of this controversy is that some people like artists, and want them to be financially supported even if this results in worse and more expensive art.
For the hell of it, a few months ago, I decided to try an AI art generator. I went with Bing Creator, since it seems to be well-regarded and it’s free. The above picture is a World War 2 biplane flying over Medieval Tallinn (capital on Estonia) in a pop art style.
The image quality varies a lot. A fan of Spider-Man, Sonic the Hedgehog and New York City, I asked for pictures of Spider-Man chasing Sonic the Hedgehog at Coney Island. Bing Creator approximates Spider-Man consistently, but Sonic not so much.
For this one, it kind of made a Spider-Man with Sonic’s mouth. And Sonic is certainly not on-brand.
This was a nunnery with three people in a forest. It came out okay in one image, even if it got the number of nuns wrong.
This nunnery seems unholy.
These poor sisters appear deformed.
Film language is versatile, so it is possible that there will be work-arounds to the shortcomings.
This short is a silent film based on shapes more than a hundred years ago. People adjust to things.
What Would Spidey Do? 