#there are a lot of arguments against generative ai which I'm not even touching

[backslashdelta]

I like to think that if you're in a fandom space posting AI art, it's because you just aren't aware of why that might be a problem.

So in case you aren't aware: the generative AI you might be using to create "fanart" is trained on a countless number of artworks created by artists whose permission was not asked. Or maybe it was, technically, intentionally buried in a terms of service document somewhere, or in an update which they could either accept or else delete an entire social media account with years of history. This "art" is only possible because of stolen art. When you post images created by generative AI, you are supporting that.

Also, fandom is about creation and community. Individuals and groups creating art of all sorts, and sharing it with each other. When you post AI art... you haven't really created anything. To me at least, that's kind of detracting from what fandom is about. I want to share and raise up works of art that my community has worked hard to create, not something generative AI spit out from a couple key word inputs.

I can't tell anybody here what to do. If you are posting AI art, I'm sure there are some people who will like it, who will share it, who will be happy to do so. But there will also be people who aren't happy about it. There will be people – and especially artists – who will be frustrated to see other people in their community who support this technology that has stolen directly from them. There will be people who will block you, who will refuse to interact with your posts, and it will make your community smaller.

I understand it's tempting, especially if you aren't an artist yourself. But if you want art that fits a specific prompt, there are other ways to get it that actually support the fandom community: submit your prompt to a fanartist who accepts prompts, or commission someone for a piece of art, or even give it a try yourself and start learning to actually make art!

I don't hate AI as a whole. I think there are a lot of really amazing things we can do with AI, if it's used correctly. But posting AI fanart is not one of them.

[deepdreamnights]

The Age Old Debate: Fire Good, or Fire Bad?

This was originally going to be part of this thread, but the points were distinct enough and my thoughts rambly enough that I split it into two posts.

From the recent PalWorld thread:

We gotta handle that last tag in two parts.

Part 1 "the devs admitted to using AI art to make the pals"

First off, that isn't true near as I can tell. I can't find anything of the PalWorld Devs admitting they used AI for PalWorld designs. Palworld had demo footage with Pals in it 2 years ago on June 6 with their announcement trailer, which means they would have had to have started dev much earlier than that.

This is what AI art from June of 2022 looked like:

On the left, Hieronymus Bosch's Pokemon, on the right, Charmander on Gumby.

I did a much deeper breakdown of the "used AI" accusation here. It does not hold water.

Now, I could change my mind on this point if there were linked evidence to the creators of Palworld saying this. But there isn't.

Because the accusation is repeated in a tag, there's no way to include supporting information, or even to easily directly ask the accuser for it. Many people are going to see it, internalize it, and then repeat it uncritically, and that's how rumors and witchhunts start.

Because I've seen a lot of accusations about PalWorld stealing fakemon, and I'm yet to see a smoking gun. There's barely smoke.

Gonna hit the second point in that tag, but while we're on the theme of spreading misinfo:

Part 2 of the Tag: Using AI to Brainstorm is "Bad"

This is also an assertion that would require support, and I believe it to be wholly incorrect.

Plagiarism happens at publication. Not at inception, not inspiration, not even at the production level. The only measure of whether something is or is not "stolen art" is whether what comes out at the end replicates, with insufficient transformation, an existing, fixed expression. Art theft is about what comes out, not about what goes in.

For more about how this works with AI art, I suggest checking out the Electronic Frontier Foundation's statement on the issue. They're the ones looking out for your online civil rights, and I agree with their position on this.

The argument that AI art is theft because it is trained on public-facing material on the internet just doesn't fly. Those are all fixed published works subject to inspiration, study, and transformative recreation under fair use. The utilization of mechanical apparatus does not change that principle.

And fair use that requires permission isn't fair use. That's a license.

Moreover, altering the process to put infringement at inspiration/input or allowing the copyrighting of styles would be the end of art as we know it.

There's no coincidence that the main legal push against AI art on copyright grounds is backed by Adobe and Disney. Adobe is already using AI art as a pretext to lobby congress to let them copyright styles, and Disney owns enough material on its own to produce a dataset that would let them do all the AI they'd ever need to, entirely with material they "own." And they're DOING THAT.

The genie is out of the bottle, they (Disney, Adobe, Warner Bros, Universal) have it, and it can't be taken away from them. They just don't want anyone else using AI to compete with them.

Palworld didn't use AI to conceive of its critters. If it had, they'd have probably been less derivative.

(three random AI fakemon I prompted up as examples of just that)

Both traditional and AI-assisted art can plagerize or be original, its entirely based upon how the techniques are used.

Moreover, you can infringe entirely accidentally without realizing, but you can also fail at copying enough that it becomes a new protected work.

We're well into moral panic territory with AI in general, and there's more than a touch of it around Palworld, largely because people aren't suspicious enough of information that confirms their worldview.

I used the quoted set of tags as the prompt for the top of the post, all the AI images in this post are unmodified and were not extensively guided, and thus do not meet the minimal expression threshold and should be considered in the public domain.