This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.
Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.
We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:
-
Shaming.
-
Attempting to 'build consensus' or enforce ideological conformity.
-
Making sweeping generalizations to vilify a group you dislike.
-
Recruiting for a cause.
-
Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.
In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:
-
Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.
-
Be as precise and charitable as you can. Don't paraphrase unflatteringly.
-
Don't imply that someone said something they did not say, even if you think it follows from what they said.
-
Write like everyone is reading and you want them to be included in the discussion.
On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.
Jump in the discussion.
No email address required.
Notes -
I think over the last few months we've established that AI issues are on topic for the culture war thread, at least when they intersect with explicitly cultural domains like art. So I hope it's ok that I write this here. Feel free to delete if not.
NovelAI's anime model was released today, and it's pretty god damned impressive. If you haven't seen what it can do yet, feel free to check out the /hdg/ threads on /h/ for some NSFW examples.
Not everyone is happy though; AI art has attracted the attention of at least one member of congress, among several other public and private entities:
I don't really bet on there being any serious legal liability for Stability.AI or anyone else, but, you never know.
I've tried several times to articulate here why I find AI art to be so upsetting. I get the feeling that many people here haven't been very receptive to my views. Partially that's my fault for being a bad rhetorician, but partially I think it's because I'm arguing from the standpoint of a certain set of terminal values which are not widely shared. I'd like to try laying out my case one more time, using some hopefully more down-to-earth considerations which will be easier to appreciate. If you already disagree with me, I certainly don't expect you to be moved by my views - I just hope that you'll find them to be coherent, that it seems like the sort of thing that a reasonable person could believe.
Essentially the crux of the matter is, to borrow a phrase from crypto, "proof of work". There are many activities and products that are valuable, partially or in whole, due to the amount of time and effort that goes into them. I don't think it's hard to generate examples. Consider weight lifting competitions - certainly there's nothing useful about repeatedly lifting a pile of metal bricks, nor does the activity itself have any real aesthetic or social value. The value that participants and spectators derive from the activity is purely a function of the amount of human effort and exertion that goes into the activity. Having a machine lift the weights instead would be quite beside the point, and it would impress no one.
For me personally, AI art has brought into sharp relief just how much I value the effort and exertion that goes into the production of art. Works of art are rather convenient (and beautiful) proof of work tokens. First someone had to learn how to draw, and then they had to take time out of their day and say, I'm going to draw this thing in particular, I'm going to dedicate my finite time and energy to this activity and this particular subject matter rather than anything else. I like that. I like when people dedicate themselves to something, even at significant personal cost. I like having my environment filled with little monuments to struggle and self-sacrifice, just like how people enjoy the fact that someone out there has climbed Mt. Everest, even though it serves no real purpose. Every work of art is like a miniature Mt. Everest.
Or at least it was. AI art changes the equation in a way that's impossible to ignore - it affects my perception of all works of art because now I am much less certain of the provenance of each work*. There is now a fast and convenient way of cheating the proof of work system. I look at a lot of anime art - a lot of it is admittedly very derivative and repetitive, and it tends to all blend together after a while. But in the pre-AI era, I could at least find value in each individual illustration in the fact that it represented the concrete results of someone's time and effort. There are of course edge cases - we have always had tracing, photobashing, and other ways of "cheating". But you could still assume that the average illustration you saw was the result of a concrete investment of time and effort. Now that is no longer the case. Any illustration I see could just as easily be one from the infinite sea of AI art - why should I spend any time looking at it, pondering it, wondering about the story behind it? I am now very uncertain as to whether it has any value at all.
It's a bit like discovering that every video game speedrun video you see has a 50% chance of being a deepfake. Would you be as likely to watch speedrunning videos? I wouldn't. They only have value if they're the result of an actual investment of time by a human player - otherwise, they're worthless. Or, to take another very timely example, the Carlsen-Niemann cheating scandal currently rocking the world of chess. Chess is an illustrative example to look at, because it's a domain where everyone is acutely aware of the dangers of a situation where you can't tell the difference between an unaided human and a human using AI assistance. Many people have remarked that chess is "dead" if they can't find a way to implement effective anti-cheating measures that will prevent people from consulting engines during a game. People want to see two humans play against each other, not two computers.
To be clear, I'm not saying that the effort that went into a work of art is the only thing that matters. I also place great value on the intrinsic and perceptual properties of a work of art. I see myself as having a holistic view where I value both the intrinsic properties of the work, and the extrinsic, context-dependent properties related to the work's provenance, production, intention, etc.
TL;DR - I used to be able to look at every work of art and go "damn someone made that, that's really cool", now I can't do that, which makes every interaction I have with art that much worse, and by extension it makes my life worse.
*(I'm speaking for convenience here as if AI had already supplanted human artists. As I write this post, it still has limitations, and there are still many illustrations that are unmistakably of human origin. But frankly, given how fast the new image models are advancing, I don't know how much longer that will be the case.)
EDIT: Unfortunately, this dropped the day after I wrote my post, so I didn't get a chance to comment on it originally. Based on continually accumulating evidence, I may have to retract my original prediction that opposition to AI art was going to be a more right-coded position. Perhaps there are not as many aesthetes in the dissident right as I thought.
I am old enough to have developed physical film in a real darkroom. Using negatives and developer was real work that took skill and helped build an appreciation for film photography. I think it was a really fun thing to learn, and I'm glad a did it, but oh boy is it useless now. Now, I can pick up my digital camera and have it automatically focus, adjust settings and snap off pictures at ~12 fps, apply all of the lens and color corrections and spit out gigabytes of jpgs onto a tiny memory card, then I take all of those photos and store them on my multiple-terabyte hard drive with backups in several locations.
I don't think the new-found ease in photography has rendered it inherently cheap, but has certainly opened the floodgates to a morass of shitty, low-level photography. I shudder every time I see a 'gram-girl (or boy) taking some basic selfie at a scenic location. But there is still a lot of photography to appreciate, pictures that take real work, not just in getting the right shot but in setting up the camera even now. No matter how smart the camera itself is, you still have to be at the location and looking the right way at the right time, and no matter how good it is at selecting a generically good setting itself, a skilled human can do better.
I installed Stable Diffusion a few days ago and, let me tell you, it's the real deal. My dumb, artistically-challenged caveman brain can put in 75 characters or less of generic prompts and in just a few minutes select from a slew of reasonably decent AI-generated art, select one and spend an hour refining it down to something I really like. It's the real reason I will be upgrading my several-generations old video card when the new 40XX series drops, not my habitual gaming, so that I can speed up that generation process. It’s absolutely blowing my mind, and I find it so very exciting to think about how I’ll be applying it to RPG or writing art.
But it’s still not magic. It has trouble taking very specific commands, it has trouble with anatomy, it has trouble with some prompts, it’s still limited in how many prompts it can handle, etc. It has a lot of limitations, many of which will most certainly go away with time, but for now I would liken it to having a decent artist who will immediately draw some art for you, but you can only communicate with them via tweets (less than that even!). Much of the skill in using it comes in through using other programs to clean up the images, removing artifacts and dropping the right “seeds” of implanted features for the program to take up. Another huge part is in getting a better feeling for how to give prompts and adjust settings to really get the most out of it. I expect the skill floor to raise up over time, so yeah, we’ll be inundated with reasonably good generated art.
This is already the case though! There are people who post on Imgur just dumps of elf art or pixiv manga art, endless seas of generic fantasy concept art, so much dross that fills DeviantArt with human-made but utterly indistinguishable work, work that people have pored so much time and effort into. Out of all that, only a few gems seem worthy of to keep around. How much worse can it get? I don’t see putting in prompts as really that terribly different now from entering terms in a search bar; that a human drew every line in one and not the other feels totally irrelevant.
I think the deeper question is a feeling about how much the AI is actually creating art, and how much of it is just "it got trained on a zillion images and it's just cutting and pasting according to your prompts; you tell it you want a buxom blonde woman in a bikini sitting on a beach and it selects out of all the stored images of buxom blonde women, beaches, and bikinis and trims it as you refine your prompt".
I think the fears and opposition come from a place where it's "a human imagines the work, puts it together, creates something new" while the machine (so far) isn't creating anything because it doesn't have a mind to think, it just does as it is told. Collage art, cutting out images created by humans and sticking them together in the combination you - not even it, itself, but the human prompter - tells it to do.
Do you get what I mean?
My understanding of the mechanics of Stable Diffusion is very limited, but I don't think this "collage art" model is quite right. The computer doesn't really pull out whole chunks of images; it doesn't know what blonde, buxom or even woman are. But what it does have are statistical relations, so when it generates a bunch of noise it pulls out lines and shapes and colors based on those statistics, depending on the prompts, then makes a bit more noise on that drawing and draws again, and eventually it pulls a random-ish image from the noise. This reminds me of when I used to sketch, and I would lightly draw lines in pencil, then as the concept firms up you make your lines darker, until you're left with a fixed image that you can commit to pen. But that's all mechanical skill really; I can imagine scenes that I would never be able to sketch, much less bring to full art; the imagining and the art-drawing are separate to some degree. Is this better or worse? I don't know. But pencils and photoshop don't think either, and no one seems to mind.
But what has really made me ponderful is that the way Stable Diffusion creates art feels similar to the way I create art, and it appears to think somewhat how I think. Perhaps that's why its unrealistic mistakes go unnoticed sometimes, my mind fell into the same trap that it did and e.g. overlooked an extra finger, because my mind doesn't sit there and count fingers and neither does Stable Diffusion! It just takes what it sees and roughly maps the shape and position into the "hand" map and calls it a day. But even human artists have trouble with hands!
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link