Alex Vranas @Alex@vran.as

@esvrld@octodon.social this is interesting input. It definitely would be an issue that certain niche art styles in the corpus could end up zeroing in non-trivially on a specific artist's style. I hadn't fully considered this, and have been assuming that in the absence of an artist's name in the prompt that the art would be more generic. This immediately has me feeling more skeptical of third party models that don't share their datasets and could be weighted in such a way to be infringing by default.

@stux These are really fun! I'll make sure to clone the git repo so I too can own one of these "NFTs".

Just curious, are the backgrounds generated from Skybox AI (Blockade Labs), or sourced from somewhere else?

And, sorry if this is long-winded I just want to get back around to the topic of generative art.

What if you had never seen a cat before (unrealistic, I know). Now imagine that you look at a handful of drawings, paintings, and photographs of cats to learn what one is.

You then draw a picture of a cat, and your only frame of reference is the small selection of other people's works.

Do you owe credit to the artists and photographers who taught you what a cat looks like?

I would also argue that most popular/usable LLMs are trained on such a huge data set, that the odds of it leveraging only a single source of information in a response is basically zero. The pretense of an LLM "using your script to answer a question" isn't very accurate, as it wouldn't even have your script retained in a way that it could reproduce it, even if asked.

That said, there are new LLMs that can cite sources, and referencing other written works has never required licensing/royalties.

Let's say I write a self-help book, and somebody who read my book ends up committing one of my lessons to heart.

They then write a memoir, and then end up paraphrasing something I wrote in my book, but they only credit it as "something I learned from a self-help book years ago", without crediting me.

Would they be infringing on my self-help book, or would it be fair use to re-communicate something I said if they believed in it?

I argue fair use.

And why is it different if an LLM does this?

@greycat I don't think that comparing intellectual property rights to financial securities is a great argument coming from someone who identifies as a communist. Marx was pretty clear about the risk of exchange value superceding use value (commodification).

In your comparison, "stealing from every bank" = "borrowing from every artist". While the former could represent systemic theft or fraud, the latter, IMHO, clearly represents fair use.

@jmaris A little bit off topic, but interactions like ours' are a really good example of why I love the Fediverse. We disagreed, I admittedly got a be terse in response, but we end up still having a meaningful conversation.

This would never happen on X, or Threads, or Bluesky. We'd be tearing each other's heads off.

Just followed you back. I never mind a difference of opinion, and you are ultimately right, opinions on this subject are extremely dependent on personal context.

@jmaris I can accept that. No bad blood here, I am happy you shared your opinion, I was just a little on guard since you opened with a rather confrontational "what you fail to see here" statement that felt more like a scold than a conversation.

@jmaris I agree, being a dogmatist is a bad move, which is why I avoid stating my opinions as fact, and making condescending statements to people I disagree with as if my opinion were the only right one.

@jmaris That is entirely your own opinion. You are free to believe that "AI art has no positive impact", but that isn't a statement of fact. I'm not "failing to see" anything here, I just don't agree with you.

Personally, I see lots of positive impacts generative AI could have. Generative art has been a passion of mine since well before the AI era, and I never devalued it. I don't see the reason to start just because the new tech is especially profound.

But, again, I don't fall for the "All AI art is art theft" argument, because it just doesn't make sense.

If I type "a moonlit beach, painting by [living artist]", It could be argued as infringing on that artist, but If I just type "a moonlight beach, oil painting", who does it infringe on? Every artist the model was trained on?

If a work infringes on every artist at once, it's safe to say that it doesn't infringe on any artist at all.

I also agree that copyright law, as currently interpreted, does not support the copyrighting of AI art. I enjoy playing around with Stable Diffusion, and I would say that I have gotten pretty good at it, but I don't take ownership of the images that I generate, and I don't believe that I should be allowed to legally claim ownership over something that wasn't actually created by me.

But I still find the technology impressive and useful.

Not to say that I don't understand people's arguments. I think that being able to type a living artist's name into a prompt and emulate their style could represent an infringement, but there needs to be a sense of nuance. I don't think that training models on copyrighted materials is infringing at all. The data a model was trained on doesn't exist inside the model, so it doesn't really count as a copy.

And prompts that don't evoke a specific artist don't produce works that infringe on anyone.