Being a bit hand-wavy with it: It’s akin to torrenting music/movies. The torrented files are lossy compressed representations of the original waveform from the music producer. Limewire, or Pirate Bay, or whatever provide interface to retrieve them (download or stream). The model weights are a form of lossy compression, and inference is like a document retrieval.
One may say, “it’s like an employee working at company X, then going to work at company Y, they retain their knowledge and experience.” I would say it’s more like, employee going from X to Y, but retaining audio and video recordings of all interactions he had, notes, documents, and other proprietary info and bringing it to company Y.
Personally this whole llm debate about copyright is quite funny. As someone who very much has skin in the game(my art is trained on midjourney.), and who runs in a circle of artists, it’s interesting to see people’s ego’s come at play here. The ones who are excited about these as tools are the ones who are openly inspired and want to inspire however the ones who claim copyright infringement seem to come off as insecure, almost like they are afraid that this idea of theirs will be the last great idea they have. There’s already a separation happening in the art world of people who are exploding in creative output vs the people who are so defensive and cling to the old way of doing things.
If I had my way, I’d see copyright laws abolished completely. A complete free for all in innovation. And people who claim that without parents and copyright then there’s no incentive to make money seriously underestimate humans and their ego to continually innovate.
FWIW, humans certainly can infringe other peoples' copyrights and can do so even if they aren't actively intending to do so. There is some boundary across which you are no longer just learning something and you are now copying, and it isn't clear at all that these generative AI techniques are actively considering the latter the way a human is required to.
But, sure: if you are against the idea of copyright entirely then it is hard to consider the idea inconsistent, though I would think a world without copyright would be a particularly hard one for an artist to make money at all...
Surely you're not suggesting that there's no such thing as "original work". The production of which may have very high capital and labour costs - which if not protected from theft - would remove the incentives of producing original work.
>As someone who very much has skin in the game(my art is trained on midjourney)
I don't know your specific situation, but there's obviously different scales of importance here. What if your art was your sole source of income, and people were reproducing it under their own name? or if you had a product where you poured millions into developing some novel IP/methods, and some employee brought it with them when they went to work at your competitors?
> some employee brought it with them when they went to work at your competitors?
Other programming languages have copied lots of D features. We at the DLF don't mind at all. Though often they copy them and kinda miss the mark.
(Yes, we sometimes copy features from other languages, too, and try to improve on them.)
This concept has specific technical meaning -
https://www.nolo.com/legal-encyclopedia/fair-use-what-transf...
It seems obvious to me that to call model weights "lossy compression" is not only incorrect from a technical (software dev) point of view, but also from this legal perspective.
The weights serve a different purpose than the original works from which they are derived, and wouldn't/couldn't POSSIBLY exist were it not for the original work of the authors of the models.
It's bad practice to go around espousing strong and condemnatory opinions about topics you don't have a full grasp of. In this case, it's both the technical details and the legal system.
It makes you look like a fool and costs you your credibility amongst peers in future encounters.
Your arguments regarding "transformative work" is what the discussion is all about. Let's see where it lands with case law.
FWIW, my stance is copyrighted content should not be used in training without request.
>It's bad practice to go around espousing strong and condemnatory opinions about topics you don't have a full grasp of. In this case, it's both the technical details and the legal system. >It makes you look like a fool and costs you your credibility amongst peers in future encounters.
Agree, thank you for the feedback. My analogy was quite exaggerative.
I'm certainly no expert on copyright law, but my understanding is that its purpose is to protect the financial interests of certain creators from the progress of technology (e.g. copy paste). I've heard arguments that removing copyright would lead to less creativity or reduced quantity or quality of work, but I'm personally a bit skeptical (probably for the same reasons as you - I think people have a natural desire to create). Even in terms of financials, I would speculate that an employment/patronage model would become more widespread.
I think there's something to be said about the benefits of having freely available knowledge, music, and art for common consumption. When I was a child in high school (or well, always lol), my parents couldn't afford a lot of material I needed or wanted for studying (especially for standardized testing, SAT and AP tests) and most of the books in my local library either did not exist or were outdated. But when I discovered that much of this information could be found online, it really changed my world and made success in life feel attainable to me. I consider myself quite wealthy now, but I don't think I would have been able to escape poverty if all this information was paywalled from me. Maybe others would argue the writers are not being compensated for their efforts, but if there are other people in the world in the same position as past me who could positively benefit from it, I think that's a better world to live in, personally.
Incidentally, the release of StableDiffusion has actually inspired me to draw a little. Not sure why, but I find it inspiring being able to iterate on a prompt and produce something of quality that I can try to replicate on my own. Even if I fail, I still have something to appreciate that maps fairly well to the concept in my head.
My hope is that these technologies might lead to a change in our financial system (I think UBI would be a good idea), but I suppose we'll see where everything ends up. I think there's likely going to be a lot of pain in the short-term (especially since there are those who don't want to adapt), but hopefully everyone will positively benefit in the long-term.
I disagree, the internet wouldn't be half as full of knowledge as it is if it weren't for the loudly ignorant giving the experts somebody to correct.
When someone teaches me, they don't own all my future creative output.
When someone teaches an AI, they do.
That's the principal difference between human learning and machine learning.
Yeah yeah, your side are the good guys and the other side is a bit dodgy.