It seems like your faith is much higher than mine that people are vetting the AI tools they use, or that they exclusively use their own works as training material.

Here vs other instances? Absolutely, yes, more would be. That said, that isn't what I said - I said they exist, and that people here would predominantly be running locally. Personally I train my own models, but I don't presume everyone else is. That doesn't change the core of the issue in your comments - the complaints are about capitalism, genai is just the context for that.

From what I can tell, our stable diffusion art communities make no distinction between training sets, nor do they require that shared images be trained on public-domain or user-owned data only. Given that, I don’t think it’s completely unreasonable that people are equating stable diffusion users with users generating their content on the big models that were indiscriminately fed the entire internet. There’s no way to easily tell.

There mostly aren't ways to tell, so there would be no requirement of that. Add to that this being an instance that would support taking the models from corporations and not using their systems to run them and that being taking away from corporations, yeah, I'd bet some do.

That doesn't really change capitalism being the root of the problem though, does it?

Here are a few of the topics I think need to be examined more, both by human society at large, and by AI-art communities especially:

• What does “good artists borrow, great artists steal” mean when the artist in question is modulating their output by inhuman means - parsing millions of images in ways that are physical impossibility? I think that’s worth interrogating.
• What say do living artists get in who uses their work in training sets, and how should that be respected? Is ignorance of publicly-stated wishes an acceptable excuse? How should this be moderated?
• How do we assign value (cultural, economic, personal, sentimental, or any other) to creative works? I think arguably that both human-created and generative AI art are the product of thousands of years of human creative output, but they’re vastly different in terms of the skill, types of knowledge, and time required to create one piece.

I think these are fantastic questions! I also wouldn't call them anti-ai, which is the part I was calling out. The anti-ai folks think that think any and all use of these models is somehow wrong, and then answer with an issue that at its root is just "capitalism is the problem" is what I was calling out.

And it worries me that a lot of people seem pretty inclined to dismiss criticism of AI use as frivolous or reactionary, or couch it as a base refusal to adapt or learn new technologies. Especially when the people driving policy around the largest implementations of that technology are the ones who are the least principled in its deployment.

I don't believe I said anything like that, what I did was say that bringing up something that boils down to "capitalism" doesn't make sense as a reason to bring up on this instance, and its what many anti-ai folks do. I think the questions you had above are exactly the type that should be welcomed and explored, but if someone is coming to a community hosted here to just complain about genai and downvote things because they don't like that - well, thats equally as uninformed and unhelpful.

But I worry that if we don’t have this kind of discussion here, where people are (maybe, optimistically/flatteringly) more judicious in their use of AI than elsewhere - if we don’t have clear, principled guidelines, then the prevailing attitudes are ultimately going to wind up being those of Microsoft, Google, OpenAI, or fucking Grok.

I think discussions are great!

I just think people immediately spewing off "Ugh AI" or "slop" or whatever is not productive and has no place on this instance.

For now though, unless I know that someone is using models trained on their own work, or at least public-domain works, I feel like I’m crossing a picket line, and I don’t like that.

And you absolutely can feel that way, 100%.

What would be inappropriate would be the above comments because of a complete lack of being informed that some people do train their own models, do use public domain works, etc, or complain about the non-public domain works for a reason that boils down to a persistent problem - capitalism - and just saying its genai that is the problem, not the capitalistic nonsense behind those large corporations abusing everyone's work and then selling it.

You seem to believe that all of the genai platforms are Gemini, copilot, etc.

This is decidedly untrue.

There are many models built entirely on public domain works, not made by or for the benefit of any corporation or business entity.

I have personally built models (not llm, thats not my use case) for identifying certain movement patterns in animals. I have made others to identify problems in audio.

The sampled data is all mine. There is no company backing it, no corporate overlord.

Capitalism is not involved.

In what way is it a "tool of the ownership class" for me to use my own models for my own use?

In the same vein, in what way are generative ai models, developed on readily available, public domain materials, provided equally to all possible both as the model (as well as available processing for free as you'll find here) a "tool of the ownership class"?

I'm not trying to be dismissive here, but what it sounds like to me is that you have limited knowledge of these solutions, and are suggesting all of them are owned by MS/Google/Meta/OpenAI/etc, and that isn't remotely accurate.

Thats like saying I shouldn't use a wrench I made in my metal shop at home, because Snap-On makes wrenches, so wrenches are a tool of the ownership class.