Agreed. The problem is that so many (including in this thread) argue that training AI models is no different than training humans—that a human brain inspired by what it sees is functionally the same thing.

My response to why there is still an ethical difference revolves around two arguments: scale, and profession.

Scale: AI models’ sheer image output makes them a threat to artists where other human artists are not. One artist clearly profiting off another’s style can still be inspiration, and even part of the former’s path toward their own style; however, the functional equivalent of ten thousand artists doing the same is something else entirely. The art is produced at a scale that could drown out the original artist’s work, without which such image generation wouldn’t be possible in the first place.

Profession. Those profiting from AI art, which relies on unpaid scraping of artists’s work for data sets, are not themselves artists. They are programmers, engineers, and the CEOs and stakeholders who can even afford the ridiculous capital necessary in the first place to utilize this technology at scale. The idea that this is just a “continuation of the chain of inspiration from which all artists benefit” is nonsense.

As the popular adage goes nowadays, “AI models allow wealth to access skill while forbidding skill to access wealth.”