You’re partly right that of course one side of the issue is just that the companies are undercutting the art market by offering a replacement product at prices that are impossible to compete with, but from seeing the complaints and viewpoints of artists, the copyright violation aspect of it is also a big deal to most of them. If only because someone undercutting you is already bad, someone undercutting you by stealing your own know-how and turning it against you adds insult to injury. To some extent I think people are focusing on this due to the belief that if not for the blatant copyright violations, the kind of large training sets required for powerful AI models would be economically unviable, and it’s fairly likely that they’re right (at least for now). Also, the kind of undercutting that we’re seeing with AI would be fundamentally impossible with human artists. You could have one work 16 hours a day with only bread, water and a straw mat to sleep on and they wouldn’t be productive one tenth of an AI model that can spit out a complete digital image in seconds with little more energy use than a large gaming computer. So we’re at a point where quantity becomes a quality of its own—AI art generation economy is so fundamentally removed from the human art creation market that it doesn’t just compete, it straight up takes a sledgehammer to it and then pisses on the pieces.
I also don’t think here that AI art is responding to an end user demand. Digital art is infinitely reproducible and already so abundant most people wouldn’t know what to do with it. The most critical end user application where someone might not easily find replacements for their very specific needs is, well, porn. That’s certainly one application that AI art is good for, but not one most companies explicitly monetize for image reasons. Other than that, I’d say the biggest demand that AI art satisfies is that of middlemen who need art to enhance some other project: game developers (RPG portraits, Visual Novel characters, sprites, etc), writers who want illustrations for their novels, musicians who want covers for their albums, and so on so forth. This goes all the way up to big companies who are already beginning to use AI art for movie/show posters (which honestly is just cheap on their part, since the budgets for those things are already so inflated they might as well pay a human artist and it’ll be a tiny fraction of the total costs), or that are eyeing the possibilities for animated movies (Jeffrey Katzenberg, of Shrek fame, said as much just the other day ).
You’re partly right that of course one side of the issue is just that the companies are undercutting the art market by offering a replacement product at prices that are impossible to compete with, but from seeing the complaints and viewpoints of artists, the copyright violation aspect of it is also a big deal to most of them. If only because someone undercutting you is already bad, someone undercutting you by stealing your own know-how and turning it against you adds insult to injury. To some extent I think people are focusing on this due to the belief that if not for the blatant copyright violations, the kind of large training sets required for powerful AI models would be economically unviable, and it’s fairly likely that they’re right (at least for now). Also, the kind of undercutting that we’re seeing with AI would be fundamentally impossible with human artists. You could have one work 16 hours a day with only bread, water and a straw mat to sleep on and they wouldn’t be productive one tenth of an AI model that can spit out a complete digital image in seconds with little more energy use than a large gaming computer. So we’re at a point where quantity becomes a quality of its own—AI art generation economy is so fundamentally removed from the human art creation market that it doesn’t just compete, it straight up takes a sledgehammer to it and then pisses on the pieces.
I also don’t think here that AI art is responding to an end user demand. Digital art is infinitely reproducible and already so abundant most people wouldn’t know what to do with it. The most critical end user application where someone might not easily find replacements for their very specific needs is, well, porn. That’s certainly one application that AI art is good for, but not one most companies explicitly monetize for image reasons. Other than that, I’d say the biggest demand that AI art satisfies is that of middlemen who need art to enhance some other project: game developers (RPG portraits, Visual Novel characters, sprites, etc), writers who want illustrations for their novels, musicians who want covers for their albums, and so on so forth. This goes all the way up to big companies who are already beginning to use AI art for movie/show posters (which honestly is just cheap on their part, since the budgets for those things are already so inflated they might as well pay a human artist and it’ll be a tiny fraction of the total costs), or that are eyeing the possibilities for animated movies (Jeffrey Katzenberg, of Shrek fame, said as much just the other day ).