generated images are not going to replace human art
I've been making art as a hobby for most of my life, and when the "genAI craze" really took off a few years ago, I was pretty bummed out. "What's the point of continuing to do this if some program can spit out better-looking pictures in a fraction of the time?", I asked myself.
Since then, my stance has almost flipped completely. In fact, instead of worrying that contuinuing to make art might be a "waste of time", I've become more enthusiastic about making art than I've been in 10+ years.
What happened? The most important thing I realized was how wrapped up I was in the commercialized or commodified perception of art.
Why do you make art? To have people look at it, get recognition and praise for it, and maybe even make money off of it. Duh. Of course, even back then I thought that art wasn't supposed to be like this. The goal should be expressing yourself, exploring new ideas, inspiring others, loving the process, etc. However, every time, that voice saying "but if you don't get anything from it, what's the point" crept back into my thoughts.
It's the commercial context that LLM-generated images seem (and probably are) the most dangerous in. When people talk about how those images will "replace" the work of human artists, they're not talking about fine art, or "art for art's sake." They're talking about products meant for consumption. The battle is not "AI art VS human art". It's "AI art VS commercialized art."
I've seen many people say that the problem isn't really the tech itself, but how it is used (or, if you want to sound a bit more edgy: capitalism). I agree, and I think this point is obvious to anyone who has gone even just a little beneath the surface of the discourse.
Of course, professional artists (most of which weren't exactly rolling in money beforehand) heaving to worry about their livelihoods because of image generators is a negative outcome. But "art" as an abstract entity isn't affected by that. Human art won't be replaced because it's something people like to do, even if machines can produce images more efficiently. And there will always be other people who appreciate the effort1, or the unique message filtered through a fellow human as the end point (not through a human first, and then 1000+ filters constructing the "most likely", averaged-out result based on a prompt and pictures on the internet).
Now all this is easy for me to say because I'm a hobby artist, and my livelihood doesn't depend on art. I don't mean to say that the economic implications of this technology aren't an issue, or don't have to be addressed. But realizing how art can exist and be valuable outside of an economic context caused a really strong shift in how I look at the whole thing.
For me, the main problem is deception. I remember the early days of the "AI craze", when people were playing around with the early versions of DALL-E. It was interesting, fun, and sometimes even inspiring ... because the results clearly didn't look like what a human would make.
But now, especially online, there's this constant doubt that you might get "tricked" into looking at generated imagery ... and many of the technology's proponents seem really keen on obfuscating their use of it. For every "proud AI artist"2 out there, there are ten users that juuuust put the stuff out there and try to "game the system" by making it seem like that what the machine spit out for them was handmade (and thus more - or actually - valuable). Why? Because of attention, and because of money. They too are trapped in the perception of art as a commodity, a product that is consumed and then discarded.
Maybe this is too optimistic of a take, but my hope is that the recent developments in tech cause people to re-examine what art really means to them, and if "pretty pictures" is really all it is. I want to say that for most it won't be, and artists will continue to thrive - in one way or the other.
Just like people still run marathons, and others still appreciate them for doing that, even though we now have dozens of "more efficient" ways to get from point A to point B.↩
I actually wouldn't mind this term so much if more people using the tech were upfront about their use. The human input in LLM-generated imagery gets so muddied down and averaged out that I lose interest in the result of the process, but the input is still there, so I guess it's technically art in the strict sense. If I could always tell, I could just ignore it, like a lot of other art that I personally don't care for.↩