Hacker News new | ask | show | jobs
by kibbi 886 days ago
I'd find it hard to argue against this, or the Penny Arcade's statements, since I'm having trouble understanding their concrete arguments in between the rhetoric. I'd be hesitant to even discuss this in their comment sections or social media channels.

One might ask: Under what circumstances would AI art be acceptable then?

For example, does it really matter if these models are created by large corporations? I don't see what the legal or ethical difference would be if it was an individual who created such a model.

Is it relevant whether their artworks were used in the training data? Well, what if a new model that is trained only on public domain photos, videos and artworks turns out to be just as capable? What if a future model is able to imitate an art style after seeing merely one or two examples of it?

It might just be a matter of time until such a model is developed. Would it be alright then? If not, why?

(Personally, I think it's the responsibility of the AI model user to use the AI art legally and ethically, as if the user made the image themselves.)

3 comments

Under what circumstances would AI art be acceptable then?

I hate to make a sort-of standard Internet retort but artists (and "society") don't have any obligation to reserve some space for AI art to be OK within culture. Maybe such a possibility exists and maybe it doesn't. But given that present AI is something like a complex but semi-literal average of the art works various largish companies could find, it seems reasonable to respond to people's objections to that.

And society doesn't have to keep paying artists. It's a terrible, distopian future, but there's nothing to say we can't.
Society has "underpaid" artists for millennia.
I think these are the interesting question. Especially if the user provided the context Style for samples.

It's currently fair use to give an artist paintings and say I want something like this but different in these ways.

You can tell a script writer to watch Star Wars and write something similar.

Questions of copyright will depend on if the output is sufficiently transformative, not if copyrighted work was used as inspiration

That question - "Under what circumstances would AI art be acceptable then?" - is definitely not asked enough. And I think taking time to make it acceptable is a worthwhile goal, part of the path, not an obstacle.
> "Under what circumstances would AI art be acceptable then?"

Easy!

Under the circumstances where the artists whose art was used to train the model explicitly consented to that (without coercion), licensed their art for such use, and were fairly compensated for that.

Plenty of artists would gladly paint for AI to learn from — just like stock photographers, or clip art designers, or music sample makers.

Somehow, "paying for art" isn't an idea that has entered the minds of those who use the art.

Perhaps because it wasn't asked enough.

This only makes it harder for open, non-profit models to compete with large corporations in making AI models. Examples: Adobe Firefly. Getty's AI. Probably Disney's internal AI. Are those acceptable, then? I'd say Stable Diffusion is more acceptable than any of them.

Information wants to be free, and you shouldn't need consent to reuse information. Furthermore, this isn't coercion -- I could just as easily make the case that artists are trying to use coercion (government power via copy-restriction laws) to impose restrictions on AI trainers.

I would say the better approach is to try and pirate the large models as much as possible -- using them to train smaller, open models, so we aren't forced to use commercial AI as much.

> This only makes it harder for open, non-profit models to compete with large corporations in making AI models.

This is not a good argument to not pay people. Minimum wage and labour laws make it harder to compete, but they are still a good thing for society.

We don’t let companies use unpaid labour of non consenting others in different cases just to make it easier to compete.

The argument is that trying to make more restrictive IP laws only benefits large corporations. Therefore the question is "only corporations have AI" or "everybody has AI".

For the former, once the contributors have been paid, all of the capitalist-related issues that come with AI will still happen, just worse. More generally, the issues of artificial scarcity get worse as well, since now the models are no longer an abundant resource, but one controlled and restricted for profit.

Unless the answer is just "I don't care if future people get screwed over, I just want to be paid once/receive pennies in royalty".

Using your analogy, it's like minimum wage laws prevented open source software from existing because contributors need to be paid minimum wage.

>Using your analogy, it's like minimum wage laws prevented open source software from existing because contributors need to be paid minimum wage.

I'd say it's different, either you work directly on an open source project and you know the license your work is under, or you license your own work to be/not be allowed to be used by for-profit companies.

>More generally, the issues of artificial scarcity get worse as well, since now the models are no longer an abundant resource, but one controlled and restricted for profit.

Sure but that should be done on a higher level, that takes on copyright largely instead of just letting artists and writers be screwed over. We are currently in a situation that only benefits one party anyway.

information wants to be free in the same sense that hydrogen does: it tends to leak. it's a ~physical property of information, not an ethical stance.
>This only makes it harder for open, non-profit models to compete with large corporations in making AI models.

No. This only forces open, non-profit models to think a bit harder about incentives to provide artists for contributing their work into the training set.

What you exhibit here is a failure of imagination.