A Few Thoughts on AI-Generated Art

The "soul" of art

The true humanity of art, I believe, lies within the effort of thinking, planning and executing the piece rather than the piece itself. When people make art (or anything) with their own prowess, they, by paying efforts, metaphysically put a part of themselves into their art, and thus an approval on AI-generated art is at the same time an indirect denial of the very existence of the artists; it's no wonder why the majority of (at least a good part of) artists hated the idea so much, and why the people who supports and "makes" AI-generated art seems to all be obnoxious people: it's not about making the artists lose their working opportunities et cetera, it's about the fact that how scarily easy it is for people who aren't willing to pay effort as a human to actually make other people believe that paying effort does not make a difference and is thus meaningless. It's not the fact that "AI making art" is scary, but the fact that "human can simply become non-human and equates to non-human" is.

By "a part of themselves", I didn't mean "human effort", but rather "unrepeatable human effort", i.e. the effort that will not humanly be the same even if the one who made it originally tries to make it again. For example, the toy that your grandma made for you when you were 7 would be humanly different from the one she made when you were 14, even in terms of it itself they are one and the same, and both toys would certainly be humanly different from the ones mass-produced in a factory. Unrepeatable human efforts are almost always made towards a specific man-made purpose, so the rejection of all this is almost sounds existential nihilistic; yet existential nihilism does not care about who you are. If we are really talking about meaninglessness, AI does not turn non-artists into artists, it only proves that making art - paying effort to make art - is fundamentally meaningless. It's less of an "improvement on productivity" than a yet-again-repeated statement about the existential situation we're in.

The glory of art comes from the struggling. Whether that struggling is meaningful or not is to each their own, but if there's no effort, there's no glory to speak of. But then of course, whether people should sink that low to shamelessly take pride in things that they have no parts in is also to each their own...

The "copyright" of art

Image generation AI, in my opinion, should be best seen as an artist that does not need to eat or sleep but only costs computing resources, and thus the act of using AI to generate art should be seen as a commission; the copyright of the generated art, if there is no consent (and exactly because the current AI is not able to grant any consent), belongs not to the person writing the prompts, not to the people who created the AI, but to the AI itself; its ownership of the art it created is just as valid as the tree's ownership of itself and the land around it. Thus from then on, the AI itself and the people who provided the dataset used for training without the data's original owner's consent is infringing the copyright of the original artists, and the people who claimed AI-generated art as their own for the purposes they have intended are responsible for infringing the copyrights of the AI.

For a lot of problems, it's better to consider AI an entity in its own right (and its creators its legal guardians). It's actually not too bizzare of a concept, the concept of "entity as tools" have existed for time as long as the human civilization itself: for example, on a farm, cows and horses are used as tools for ploughing; in a company, employees are used as tools for bringing in profits. The notion of "machine rights" really isn't that far away from us if we keep going down this path, and sooner or later we will have to accept the fact that our machines are instructed by some people to unjustly rob other people and the same people are unjustly robbing our machines right now...

The "future" of art

Art will not die, at least I don't think so, no matter how loud the "pro-AI" people yells: I have yet to see an AI-generated artwork that has deviated enough from previously existing styles, which means instead of AI, it's still human who retains the ability to create new styles. I also don't think AI can make modern art like The Art Guys, at least in the next decade; and if that ever happens, it means that AI understands concepts just as well as humans do, and they making good anime waifu artworks would be the last thing humanity should be concerned about...

I'm not particularly worried about AI-generated music either, as two things could have happened: (1) some people are forced to make (and listen to) more "extreme" music, which is kinda already the case in the 20th century and (2) some people might just not listen to music anymore. Thanks to western style pop being the dominating music style of now, only a rather restrictive part of music is going to be "lost to the hand of AI".

Automatically generating songs have been long possible before any machine learning AI comes into the sight of the general public; I have played with chord progression generating tools (with different orchestration) a lot as a kid already, music did not die back then, and I'm certain it won't die in the next decade. We have been summing up the rules of music for centuries so precise that one can already code them into a classical rule-based AI because teaching the rules (and how to break them) is how we teach young musicians, and I doubt any of them AI is capable of producing anything close to what The Mars Volta had been doing or the Bach drop in Procol Harum's Repent Walpurgis...