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 -
Well, most of what professional commercial artists do on a day-to-day basis isn't exactly creative either.
The paradigmatic examples of creativity are novel ideas that register on a world-historical scale, such as Cantor's development of set theory and the hierarchy of infinite cardinals. Such ideas are necessarily rare. If a work doesn't fit into this elite class, then I question if it can be called genuinely creative. At times I have wondered (but never seriously believed) if any work of art could ever be novel enough to qualify as genuinely creativity. As Hilbert once quipped, "for a mathematician, he did not have enough imagination, but now he has become a poet, and everything is fine."
I don't want to position myself as the god-emperor of creativity, or pretend that I have a set of hard and fast rules to apply. I just want us to have standards for ourselves, is all. As we descend further down the scale towards ordinariness, from world-historical successes, to works that are widely considered to be of exceptional quality, down to the average things that average professionals produce in their average careers, it becomes less and less clear whether the adjective "creative" continues to apply. I am not proclaiming anything with certainty one way or the other. There is just less clarity.
AI art obscures these questions and pushes them away from the central place that they should occupy in our thought.
"Wheee yippee, now we can all be creative! Thanks, AI!"
I find this to be offensive nonsense.
How would you compress this into a prompt? Without using explicit identifying terms like "Paul Klee" or "Angelus Novus".
I encourage you to run your prompt by the AI and see how close it can get.
This seems like an extremely unusual definition of creativity, one that, by your own admission, excludes most and perhaps all artists, and one that neither I nor most others share.
I don't really know or care whether current AI has this level of "Ur-Creativity". It is sufficient for my purposes to note that to the extent that human artists engage in creativity, the AI does as well. If you feel that the AI moves us further away from "Ur-Creativity", you should feel the same about the pursuit of art by humans as well. What you don't have, I argue, is reasonable grounds for treating one different from the other.
Standards that, by your own admission, ~0% artists are now or have ever met. Why should I value such standards? What if I don't believe that art can "register on a world-historical scale", or particularly care if it does? Compare this to a definition like "the ability to produce novel constructs that please and satisfy other humans".
I don't think there is, actually. You're noticing that, as the sage says, there's nothing new under the sun. And that's true! but it doesn't actually matter, because things can be new to you, and humans care about the details. By one accounting, creativity is impossible, because everything we make, boiled down far enough, is just a rehash of a remix of basic themes depicted a million times before. By another accounting, every 5-cent American flag postage stamp is perfectly unique. The truth is somewhere in the middle, and it's somewhere in the middle because our brains actually adjust the degree of novelty they're searching for, based on other inputs. At the end of the day, you haven't presented an argument for why normal creativity is a bad thing; you've just asserted that you can sort of imagine something better, in the abstract, maybe.
It doesn't, any more than existing artists and existing art tools have. The only new thing it does is be fast, tireless, and free. To the extent that you have a problem with this, I think you have a problem with art itself.
It's said that there's two points in the timeline of any AI project: the time when it's better than a human for the first time, and the point at which any human is better than it for the last time. We've crossed the first of these points, and the second will likely be reached within our lifetimes. Art is not magical. It is a thing we do with our brains, it's a thing we can learn and teach, and that means it's a thing we can automate. Everything valuable you derive from the visual data of a given piece of artwork can be described, analyzed, reduced to a formula, and reproduced via other humans or by an AI tool. There is no secret sauce, no higher mysteries. We are already starting to do the art version of the Turing test, people are frequently losing that test, and you will not do better.
By analyzing the elements used in the piece, which are actually not terribly complicated. line weight and quality, the abstract treatment and the heuristics used to describe the figure's form and the various surfaces it contains, the figure/ground inversion games... It uses a technique similar to one of my favorite artists, a sort of forced contour simplicity that trades off distortion of form for a sort of internal balance of composition. All that on top of the specifics of texture and color. I'm speaking in generalities here because the piece is idiosyncratic enough that common vocabulary hasn't been developed and disseminated sufficiently to do otherwise, and for good reason: it looks maybe 10% interesting, and 90% like shit. I don't want to draw things like this, or particularly to look at them, but if you payed me fifty grand I could spend two weeks figuring out how to crank them out more or less on command. I'm distantly aware of Paul Klee, and I've never heard anything to make me dislike the man, but his work is not valuable to me in any significant way, and I doubt the work itself is valuable to most other people either.
I agree that it would probably be pretty difficult to get a current-gen AI to produce an analogue to that painting without using the prompts you excluded. What do you think this proves, and why?
It shows that people who outsource their artistic production to the AI will be railroaded into established patterns of thought and will be encouraged to produce things that can be expressed in familiar terms, rather than things that can't be.
I don't think it shows that at all. You asked for repetition, not novelty; you wanted a rework of a specific artist's specific painting. How does this support a claim that AIs can't deliver works of novel quality?
The AI is going to have a hard time reproducing one specific, highly idiosyncratic style, when handicapped by denial of references. That doesn't show that it can't reproduce that style with the references, and it certainly doesn't show that it can't produce other specific, highly idiosyncratic styles de novo, which it in fact excels at. I've gotten good results feeding the AIs selected portions of my favorite poetry, for instance.
None of this implies railroading of any kind, and in fact I think you have it completely backwards. Actual artists have preferences of what they want to make, and these preferences are often heavily impacted by the market as a whole. The AI does not have preferences, and it does not care about the market. It will make whatever you want, as much as you want, fine-tuned exactly how you want. It will drive more variety, not less, because it makes true novelty significantly easier, if that is actually what you desire.
But is that what you desire? Did you pick that Klee painting because it truly spoke to you on a transcendental level, irrespective of who made it or what other people thought about it? Did you come across the picture naked on a wall somewhere, devoid of relevant context, and stood transfixed at its simplicity and direct, focused purpose? Or did you see it in a book or in a museum, couched in the appropriate status signals, pomp and ceremony to tell you that this was a Serious Work of Art by a Serious Artist, hence important and valuable?
There's two very different social games that one can play with art. One involves exploration, the delight of novel recombinations of the familiar transcending into the sublime, of subtle webs of discovery and recognition. the other is naked status games. AI can probably do either about equally well. The latter, though, at least in my view, is a complete waste of everything valuable that we have.
I'm sorry if this somehow wasn't clear. The point of the experiment wasn't to think about ways the AI can reproduce a known painting that already exists. The point of the experiment was to think about how to get the AI to produce a specific image in your head that doesn't exist at all yet. You have the temporal order of events mixed up.
Suppose I am sitting down to paint the Angelus Novus, and I have an image (and I use that term very loosely - it is half an image as is commonly understood, half just an intuition that it has to be "like that", where that is not really expressible in any concrete form) in my head of what the painting is supposed to look like, but, the painting does not actually exist yet. It does not exist anywhere, no one except me knows what it should look like. I have dreamed up this painting I want to bring into existence.
Obviously I'm going to have a hard time getting the AI to produce anything that's anywhere close to what I'm actually after. If I simply ignore the AI then and draw it myself, or if I just end up giving up entirely and don't create anything, then there's no real harm done. But if I insist on using the AI anyway, and I settle for something less than I originally intended, settle for something the AI produced because it's "good enough", that's where things start to go very wrong. I may even be tricked into thinking I have genuinely created something, that I have genuinely satisfied my original vision when in fact all I have done is abandon it. That's where the real dangers lie.
I have already acknowledged in the comments here that there's a lot of nuance to unpack in this process - the original starting vision can be very indistinct, little more than phantasms, and it is subject to contingency and chance and revision in the concrete unfolding of artistic production. But I also don't doubt that people really do set out to create specific things, and that at the end they can evaluate how close they were to hitting the mark. The AI will never hit the mark in the way you could if you were capable of drawing yourself. Not even another human will.
We have already established that this is false. I gave you a specific image and asked you to produce it using the AI and you acknowledged that it could not be done.
In fact it would be easy to pull any number of images from danbooru or a random manga or whatever that could not be satisfactorily be reproduced by currently available models.
I place no value whatsoever on variety for variety's sake. It is worse to have 100 failed works of various types than to have one failed work of one type. It is of no comfort to point out that, in the 100 bad works, there is great variety.
It is one of my favorite paintings, and I find it to be very beautiful and captivating (although I picked it specifically for its unusual qualities, not merely because I find it to be beautiful).
I can't run a perfectly controlled experiment where I could show you what I would think of the painting even if it wasn't produced by a canonical artist, but I like to think of myself as being relatively unbiased. There are canonical paintings that I find to be wholly uninteresting (much of what was produced in the Renaissance), and there are non-mainstream works that I think are of exceptional quality (vaporwave is probably the most profound artistic development so far of the 21st century).
Here's the thing: every artist has probably gone through this exact dilemma, and it hasn't exactly destroyed their souls. There's that quote about making four movies: the one you conceive, the one you write, the one you shoot, the one you edit. Probably every major published writer has had the experience of butting heads with their editors. Let's not even get into video games, where literal pages upon pages of digital ink have been spilled documenting the disconnect between conception and final product.
There are probably very few people in history who could claim that they made something that conforms 100% to the visions in their mind palace. Humanity seems to get by pretty okay with imperfection. Perfection is amazing, but it has too high a price to pay, in many instances.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
How would I paint that? How would I describe it to another person such that they could accurately see it in their head? I can't do either of those things as it is. Why is it more authentic to imagine I had the painting skills than to imagine I had the prompt-making skills? I can't tell you what prompt will work, but you know, I also can't tell you step-by-step biomechanical instructions to make a real painting.
You seem to very much value the idea of art as a journey, whereas I -- and I think most people cool with this -- see art as a destination. So long as I can get pleasing outputs aligned with what I want, I can express myself artistically. It's no more art if I spend 30 years perfecting brush strokes, and it's no less art if a genie plucks the picture from my head and materializes it on a canvas.
More options
Context Copy link
More options
Context Copy link