Oh no globalcomix, not you too

Oh no globalcomix, not you too

GLOBAL COMIX AI POLICY:
https://globalcomix.com/ai-policy
What do you people think?
I'm still having a think about what to do - but I know where I'm leaning atm. It's a shame though, because GlobalComix is a revenue source for me, so I'd cut myself off from even more of my previous income and I'll be worse off.
@JenJen That reads like a pretty responsible and pragmatic policy to me.
Forbidding purely AI content but accepting that AI tools are now ubiquitous and some people will use them seems to me to be sensible given where we currently are.
It fits well with my personal opinions: I abhor outright GenAI created content, but I don't object so strongly to a bit of content aware tool usage under direction by an artist. They're unlikely to uninvent those tools, so better to lay out how they can be used.
@gilester45 What about the source material the models are trained on? I have a huge problem with them being tools for remixing other people's work, without consent, attribution or payment.
The Globalcomix statement is pointedly silent on this aspect, too.
If they were talking about purely "mechanical" assistance, like an animator using an automated system for in-betweening, I'd find their statement entirely acceptable. That even seems to be how they're casting the use of these things.
@JenJen
@KatS @feff @JenJen It's a good question isn't it.
Tools that steal our work to recreate it in our style are evil, the industrialised theft committed by AI companies beggars belief in this regard.
However a tool that has "learned" how it should work by watching our creative output... that doesn't sound so very different to how a human might learn from creators they admire.
Is that enough to say it's ok? Or, perhaps "less unacceptable" is all we can hope for.
However a tool that has "learned" how it should work by watching our creative output
If and when such a tool exists, we can also discuss its rights as a sentient being.
However, what we have now are
Tools that steal our work to recreate it in our style
They remix existing artifacts. They do not learn, and they do not create. They launder the plagiarism of other people's creations. This is why Sam Altman and everybody else driving these buses are working so hard to remove copyright protections from people who actually create.
To emphasise: people (can) learn from others, and adopt and remix each others' ideas, perspectives and techniques. These diffusion models remix existing works, and that is very much not the same kind of thing.
So no, that's not enough to say it's OK. It's not a question of how you frame it, but a question of what these things do and how they're wielded.
One exception: if you have a private, local model that is trained exclusively on your own work, have at it.