People who oppose AI-produced art aren’t angry about the existence of all types of AI. Art generation tools in particular are the subject of people’s fury, and that’s because they’re fundamentally different from a lot of other AIs.

Something like, for example, a call center AI exists to support the work of human employees. It flags up keywords to show call center agents helpful information during their calls, and can alert a manager if an ongoing call needs intervention. That’s highly useful to these employees, and instead of threatening their job security, it makes their work lives more comfortable.

AI art generators don’t have the same impact on human artists. Instead of supporting them in their creation process, AI art tools use pattern recognition tools to ‘remix’ existing work into something new. They don’t help artists figure out how best to shade that tree, for example; they can only create a new facsimile of a tree out of data from other artists’ work featuring trees.

One of the biggest problems digital artists cite with AI art generators is the issue of how they got the data they use to create new works. Or rather, the pieces they splice together, since the works in question aren’t actually created from scratch. Despite the fact that AI art doesn’t rely on breaking copyright laws, many artists feel cheated by the fact that that’s even true. Imagine, for context, that you’re part of the team behind a popular piece of software. You put in hard work to make sure the code was just right, and that all major bugs were ironed out. You’re proud of the finished product. A week later, an imitation of your work shows up on the market, retailing for half the price. The code they use looks suspiciously similar to yours … but not quite similar enough for you to be able to take them to court over it.

Did they violate copyright laws to create the imitation? No, not quite. But you’ll likely be left feeling frustrated about the lack of legal protection, instead of relieved that no rules were broken. The fact that it’s technically fair play is the problem, which is the same for digital artists.

A common theme among artists’ complaints about AI art generators is that they’re unhappy about their work being used without their permission. This is true of the vast majority, if not all, of the art generator tools available. They were trained using data scraping, without consulting the people whose work they were trained on. And since AI tools can’t generate art out of nothing, that training wasn’t passive either. It’s not the same as a human learning from Da Vinci by studying his work and trying to paint like him. If anything, it’s more like a human tracing the Mona Lisa and then claiming they made it themselves.