Hey there, sometimes I see people say that AI art is stealing real artists’ work, but I also saw someone say that AI doesn’t steal anything, does anyone know for sure? Also here’s a twitter thread by Marxist twitter user ‘Professional hog groomer’ talking about AI art: https://x.com/bidetmarxman/status/1905354832774324356
(Fair warning, I have time to do a long thing now… bear me, or don’t, up to you.)
I’d have to go find those other comments of yours, but for the moment, I will say, I kind of get it. I do remember seeing at least one comment that was sniping at anti-AI views and being uncharitable about it, and I kinda tried to just skate past that aspect of it and focus on my own read of the situation, but I probably should have addressed it because it was a kind of provocation in its own way.
But yeah, I can get defensive on this subject myself because of how often anything nuanced gets thrown out. Personally, I’ve put a lot of thought into what way and how I use generative AI and for what reasons, and one of my limits is I don’t share AI-generated images beyond a very limited outlet (I’m not sharing them on the wider internet, on websites where artists share things). Another is that I don’t use AI-generated text in things I would publish and only use it for my own development, whether that’s development as a writer or like a chatbot to talk about things, etc.
Can they be “treat generators” in a way? Yeah, I guess that’s one facet of them. But so is high speed internet in general. It’s already been the case before generative AI kicked into high gear that people can find novel stimuli online at a rate they can’t possibly “use up” all of or run out of fully because of the rate at which new stuff is being produced. The main difference in that regard is generative AI is more customizable and personal. But point is, it’s not as though it’s the only source of “easy treats”. Probably the most apt comparison to it in that way is high speed internet itself along with the endless churn of “new content”.
Furthermore, part of the reason I chose the example I did of use in my previous post is that while, yes, there are people who use generative AI for porn, or “smut” as some would call it in the case of text generation, the way you posed your post, there was essentially no way to respond to it directly without walking into a setup that makes the responder look bad. If the person says no, I don’t use it for that, you could just say, “Well I meant the people who do and I’m sure some do.” And if the person says yes, I do use it for that, you can say, “Hah, got you! That is what your position boils down to and now I’m going to shame you for use of pornography.” It also carries an implication that that one specific use would cloud someone’s judgment and other uses wouldn’t, which makes it sound like a judgment specifically about pornography that has nothing to do with AI, which is a whole other can of worms topic in itself and especially becomes a can of worms when we’re talking about “porn” that involves no real people vs. when it does (the 2nd one being where the most intense and justifiable opposition to porn usually is).
Phew. Anyway, I just wish people on either end of it would do less sniping and more investigating. They don’t have to change their views drastically as a result. Just actually working out what is going on instead of doing rude guesses would go a long way. Or at the very least, when making estimations, doing it from a standpoint of assuming relatively charitable motives instead of presenting people in a negative light.
deleted by creator
Thanks for taking the time to write a thoughtful and nuanced response about it. I don’t think I disagree with any of the concerns you brought up, tbh. They all seem like valid things to be concerned about with AI. You might be surprised to know that I know of one person who was in AI dev for years as a UX designer, including for image generation, who also had strong opinions about the value of doing creative work “by hand” (e.g. not automating it) for development, for expression, etc. May sound like a contradiction that he was like that while also working in AI, but it’s interesting to me because it helps show that not everybody who has worked in AI is a full-on tech-bro. Of course, you could argue the end result is the same, that he contributed to enabling the tech and all its consequences, but still, I find it interesting.
I do think active building/learning instead of just consuming is important. It’s a bit murky sometimes, I think, when considering things like video games, where you do have to learn, but you are also learning something that often has no application to RL. With image gen, there is learning to prompt, however minor that is, and some people do mix it with actual drawing, but overall yeah, you could low effort type in prompts and press button until you get results you want. And to make an analogy to food, that is probably more like eating sugar than nutrients. Unlike with food, you won’t get diabetes from chronically overdoing it, but you might find yourself dependent on the “hit” of the novelty and have a hard time removing it from your life. The design of image gen is very similar to gacha, even if not intentionally.
That said, in terms of self improvement as a whole and a sense of fulfillment, I’m not convinced that hobbies like drawing are a great answer in and of themselves, but they are notably better than pressing button and adjusting prompt. Where my mind goes there is, we’re looking at broader societal problems, where people consciously or unconsciously notice that the society they live in is sort of nihilistic and exploitative, and that can be depressing in itself. Individualist development of hobbies may be more of a liberal solution overall than a revolutionary one. I think we can do better, though in the interim, again the individualist solution is probably better than no solution. Similar to how chatbots can be helpful for some people for processing things, but they are no long-term solution to alienation and isolation. So there’s the harm reduction perspective on it and what form that can take, and then there’s what to aim for. I think in terms of what to aim for, we need community rebuilt and communal things for people to do more so (that don’t have big cost barriers).