AI art.

[Arioch]

I have the feeling that the word "creativity" is being used in a very broad sense here. As with "art," there's probably not much sense in arguing too much about it, but if we set aside the craft and skill of the illustrator, at a base level creativity involves the choices you make. In even the most simple illustration, there are a hundred important choices that the illustrator makes. In a comic with a script, some of them will have been already specified by the writer, but only a tiny percentage at most.

An AI package like DALL-E doesn't make choices. The operator specifies some of them, but the majority are grabbed through osmosis in the source data that it uses. So no, a AI like this can't be creative, because it's not making creative choices.

This is also why you really couldn't make a comic like Outsider with DALL-E. You could have DALL-E create an image for each panel, but even with the most detailed script and instructions, there would be little or no continuity from one panel to the next, in character or costumes or background. Each panel would be its own universe. You'd be much better off using a clipart package or something like Poser.

So no, this is not a case of an angry artist who is worried about losing his job to AI. I'm mainly irritated that people seem so oblivious to the actual capabilities of the technology (and to the legal ramifications in their indiscriminate use). I'm not interested in doing the kinds of work that DALL-E is capable of doing, and DALL-E isn't capable of doing what I do. If Outsider has any value as content, it's because of the creative choices that I make in its creation, and not through whatever questionable skills as an illustrator I possess.

And yes, at some point far enough into future, there will probably be AI's that will be much better than humans at everything. But since in that case, since everyone will be out of a job, there will be no one to hire you or pay for your comic, I don't see any reason to look too eagerly towards that eventuality. You won't be able to create a comic, because the AI will have no reason to do what you tell it; neither you nor the would-be readers have any value to offer (views and likes won't matter, because there's no point in advertising to people who have no purchasing power). I suppose our concepts of how economies are structured will have to be reinvented, but I'm not optimistic about the ideas I've heard to this point about how to run a post-scarcity "economy."

[Keklas Rekobah]

If Outsider has any value as content, it's because of the creative choices that I make in its creation, and not through whatever questionable skills as an illustrator I possess.

In my opinion, it is both. Without your illustrative skills, Outsider would look like another Randall Monroe webcomic — thoughty, yet simplistic. Your illustrative skills make Outsider seem realistic. The “line up” on the caste descriptions page is more relatable than a similar line up of stick figures.

[gaerzi]

So, you're definitely correct when you say that it's analyzing source images, however, it's not doing that in real time. It's already "digested" the hundreds of millions of images from it's training data, and it - along with other diffusion models - seem once trained able to operate without a connection to their training dataset.

I don't know if this makes a difference to your argument, as, at the end of the day, it is still analyzing patterns from word-image pairs, but I do think that should give it a slightly higher level precedence when compared to a photobasher.

And, again, you're absolutely right when you say Dall-E doesn't demonstrate understanding the same way a human does. It's said that Dall-E returns the most "probable" result for a set of words, not the "best". It does essentially function like a search engine, hence why terms like "Trending on Art Station" or "Very beautiful" can end up affecting the output so much.

But that also greatly limits the usefulness of such a thing, because essentially you can only used it to recreate art that already exists.

Like, suppose you're working on some sci-fi game and needs illustrations of your cool homebrew alien species for the game, the svinfgonians. There's no such thing as a svinfgonian on the Internet. The AI won't help you there.

Where AI can help you is if you want utterly generic assets -- and those tend to be readily available, often for free, so the AI isn't very useful either.

[Mithramuse]

This is certainly getting attention and discussion in multiple places; for one example I've run into, see "https://camestrosfelapton.wordpress.com ... mitations/". Largely seems to agree with the discussion so far, though doesn't touch on the question of where the images are coming from, exactly.

Another would be Ursula Vernon's Twitter (@UrsulaV), where she has been using various art "AI"s to make backgrounds for a comic, with commentary on what works and what doesn't. She is very much retouching the images, though, and has a storyline that purposefully allows for the variations in the images - very worth a read, both for the comic and the commentary. Last entry at present is here, at the end of one thread; the first post in this thread links back to the start of the comic.

--Mithramuse

[Keklas Rekobah]

@ARIOCH: How did you generate the 'exterior' background in the bottom panel of #217?

It looks like either a photo or a very realistic drawing.

[Arioch]

@ARIOCH: How did you generate the 'exterior' background in the bottom panel of #217?

It looks like either a photo or a very realistic drawing.

It's a 3D model with overpaint for the haze and glows.

[G. Janssen]

Dunno.. my results for
"Epic viking dolphins fighting robot hedgehogs from the moon while eating messy hotdogs"
Was a bit of a mixed bag.

Protip: ask an AI to draw itself.

[Demarquis]

None of this is anything new. I remember a learning program called Aaron that started learning how to produce representational art starting in the 1970's and ran until the programmers death 2016. IIRC, it used feedback from art critics as the basis of it's "learning".

https://en.wikipedia.org/wiki/AARON

Aaron wasn't designed to replace artists, but to shed light on human creativity: "Cohen is very careful not to claim that AARON is creative. But he does ask "If what AARON is making is not art, what is it exactly, and in what ways, other than its origin, does it differ from the 'real thing?' If it is not thinking, what exactly is it doing?"

[bunnyboy]

Protip: ask an AI to draw itself.

I dont remember exact wording, but I asked Midjourney to draw something like "Message on paper in big letters written in English."

The first one gave quite interesting view how the AI worked, as it started from clear letters, but became "mutilated" when the AI added more resolution on the image.
I thought that the language part was kind of important, as I did remember one youtube video where DALL-E2 was asked to draw two farmers with speech bubbles and the AI wrote those in its own language, which when used as prompt, always gave images of carrots and cabbages.

I also asked it tell me a joke.

[Overkill Engine]

And yes, at some point far enough into future, there will probably be AI's that will be much better than humans at everything. But since in that case, since everyone will be out of a job, there will be no one to hire you or pay for your comic, I don't see any reason to look too eagerly towards that eventuality. You won't be able to create a comic, because the AI will have no reason to do what you tell it; neither you nor the would-be readers have any value to offer (views and likes won't matter, because there's no point in advertising to people who have no purchasing power). I suppose our concepts of how economies are structured will have to be reinvented, but I'm not optimistic about the ideas I've heard to this point about how to run a post-scarcity "economy."

I foresee a lot of atrocities being a potential outcome of such a "post scarcity".

Having a huge portion of the population becoming economically obsolete means they no longer have even the minimal socioeconomic bargaining power that even the most menial laborer has now. Which means at most you have three main groups of people:

The Disenfranchised
The Automation controllers
The Government
(The last two may effectively become one and the same. Beware)

And the only way for gross abuses of power to not happen in such a scenario is to abjectly rely on the government not siding with the automation controllers that, well, control any and all resource production that matters. (And before anyone starts huffing and screaming about capitalism...such a scenario is post-capitalist. There is no longer free trade when a majority of your population has either no ability or need to trade.)


Given human history thus far I am not betting my life savings on that outcome.

[Demarquis]

Put me down as very skeptical that we can ever get to a post scarcity economy. I don't see a path from here to there.

[Arioch]

This is a very capable analysis of the legal issues behind the AI image generators and the current lawsuits against them.

[QuakeIV]

True technical post scarcity, clearly not. As my economics 100 professor put it, 'as soon as we achieve a post scarcity economy I am immediately retiring to my beach planet with my million bikini wearing concubines'.

I could see free food though, for instance.

[Demarquis]

So one question that occurs to me: can you freely "steal" the AI generated art? Or is that product copyrighted itself?

[G. Janssen]

So one question that occurs to me: can you freely "steal" the AI generated art?

Yes.

Or is that product copyrighted itself?

Copyright means that you have the right to copy it.

Can't wait to download and 3D print a Ferrari.

[Demarquis]

"Copyright means that you have the right to copy it."

If that were true, I would have received significantly better grades in school!