alt Hacker NewsThe troubles over copyright infringement in AI training data remind me a bit of Eli Whitney and the cotton gin.
There he suffered massive patent infringement, that basically stopped being enforced due to the sheer economic importance of the cotton gin.
In a similar manner, I think there is a reasonably strong argument that it was wrong to use copyrighted material for AI training without paying royalties nor even asking for permission. But equally, every country wants to have the most powerful models and enforcing such royalties would make it effectively impossible to train them as the amount of material required would cost an insane amount in royalty fees.
So I expect the law will continue to turn a blind eye (perhaps enforcing some token payments like that $1.5B mentioned in the article) because "if we don't make these models, the Chinese will" etc.
Replies
Author here. Appreciate your thoughts and I mostly agree actually.
I'll explain more over the next few essays, but I am designing my proposed regulatory structures to try to accomplish 2 purposes in tension simultaneously like the Fed: 1. Maintain global competitiveness for frontier labs 2. Create a societal hedge against the AI bull case (AKA the economic black hole case)
A % of revenue scales in a way that I think balances the two well while avoiding all the other problems I mentioned in the essay. I’ll get into ratchets, timing, and thresholds in later essays, but I agree the China/competitiveness problem is central.
> ... there is a reasonably strong argument that it was wrong to use copyrighted material for AI training without paying royalties nor even asking for permission. But equally, every country wants to have the most powerful models and enforcing such royalties would make it effectively impossible to train them as the amount of material required would cost an insane amount in royalty fees.
i think you're spot on this is one of the key arguments made beneath the surface. what i find so strikingly frustrating about it is, so many of the ai cultists [0] will imply and sometimes even outright say that writers, artists, musicians are silly useless and overvalued and the work artists do is entirely frivolous. then next breath explain why those artist's work is one of the most important things for a model to be trained on. suddenly art is very important. we absolutely must have access to their work. but also we shouldnt pay them because their work is silly and unimportant.
if an artists (musician, writer, journalist, painter, etc...) work is useless, then obviously you dont need it for training. if their work is imperative and you absolutely must use it, then pay for it.
ive noticed this with ai companies a lot. over and over again they contradict themselves to the core.
1) art is silly and not important enough to pay for but its absolutely foundational and we must be given unfettered access or our models will suck.
2) "our models are the smartest thing in the entire world. also, you're a dipshit if you trust them at all."
ill say it again, if removing art and culture from the training sets would render your model useless, then obviously pay for it.
[0] when i say cultists, im not talking about normal people who use ai. im talking about an entirely different group, we all know the types im talking about.
I'd be fine with the nuclear compromise: if AI training is allowed to infringe copyright, then there is no legal protection for the models themselves and their weights. Distillation should be explicitly legal. There will of course be a huge cat and mouse game about it, but let's have competition drive prices down on the stolen IP.