It seems like before the time AI art actually popped off, the thought of a machine being an artist was what people thought AI art would be in the future. That there would be AI musicians that could write, sing and publish their own songs, with fans admiring them and enjoying their songs, and that there would be art made by computers in museums and prize winning novels written by a machine in bookstores (like in the Roald Dahl story The Great Automatic Grammatizator).
And the thing is, at the time the whole AI mess started, you could see that this was the scenario people expected: there were AI art exhibitions, AI art winning actual art prizes, people being surprised about a [something] fully created by AI, et cetera. And meanwhile people predicted AI art would be hanging in museums in a couple years, [insert big percentage] of [insert art form] would be created by AI by [insert year], blah blah blah.
But as time went by, you could see that AI art probably wouldn’t be going in that direction. Not only were people no longer surprised by all the things AI could create, but people also started to become more aware of the mistakes image generators kept on making. It made people develop a sixth sense for when something was AI generated. The accessibility of these image generators also played a role in this. People started seeing AI art as cheap and lazy. The internet slowly (I mean quickly) became more flooded by AI content which started to annoy people looking for art made by humans. All the problems AI image generators have, like the art thieft and the impact on the climate, were of course the main reason people hated AI art, but what I just described only made it worse.
AI art became known as cheap and lazy shit used for content farming, because that was what it was mostly used for. We weren’t getting some greatly admired art machine, but instead accounts that spammed AI art and tried to profit from it. Together with the tech bros trying to convince people that AI could make movies and stuff when in reality these looked terrible and held together by duct tape, it made people feel less and less like there was any future in AI art.
This is why the predictions and things people were making for AI art in the beginning seem so strange from today’s perspective. An AI art exhibition for example now sounds like a restaurant that proudly serves microwave meals from the freezer section at the supermarket. And just like that an AI musician with a big fanbase also doesn’t sound realistic anymore, because that now sounds more like a content farm.
People simply made a wrong prediction about computers making art in the future. This is probably because image generation AI isn’t working the way people thought it would be working. This isn’t some eureka moment from computer scientists discovering how to build creativity into a machine, these are just one purpose generative AI models that just… do what you ask them to do. They don’t make an image of Ronald McDonald wearing a maid dress because they had that idea themselves, but because a human had that idea and asked them to make it. And I’m not even sure if the faith of AI art will be different if there’s a better and different AI generator in the future that does have the creativity that modern image generators lack.
If my stance on AI art wasn’t very clear from this article, I’m of the “AI art as a tool and not a replacement” opinion. I think AI can help artists get a rough sketch of an idea, but I don’t think it can replace artists (and I also don't think it should), not only because of the obvious arguments like plagiarism and large use of resources, but also because I’m not sure if the mistakes it makes can be fixed with the way they currently work and if a computer can have the creativity of a human. My prediction is that AI art will probably still stick around as a thing for content farms and deepfakers after the AI fad dies, just like how there are still crypto scams going around after the trend died out. Most of the memes created by AI were already made and the people who can see if an image is AI will probably avoid it, so those images of Jesus made out of shrimps on Facebook will probably be what it will mostly be used for.