Things are about to get worse for generative AI

It's difficult to say really.

If I essentially encode knowledge of something then can recall and remix at will, am I redistributing the exact work or the knowledge of it?

Yes, it is capable of producing a close to exact replica, if not the exact same input image byte-for-byte, but I find it difficult to say OpenAI is willfully redistributing copyrighted work in a whole like you would with torrenting a movie or right-click saving an image from Google where you are copying the intellectual property 1:1.

Opening this Pandora's box could have large implications on a lot of creative work that could cause artists to be unable to work if taken to the end conclusion: you cannot create any creative work that has a talking mouse if you have knowledge of Mickey Mouse existing because you have been tainted (similar to whiteroom re-creations but now any sufficiently large copyrighted figure causes a deadlock condition for all derivative ore even similar topics).

Is Ratatouille derative of Mickey Mouse? Ehhh, well they are both talking rodents. They both have cartoon faces. You can certainly draw parallels between them but they aren't the same character. Is Mickey with a chef hat infringing on Ratatouille?

The trademark law, to my knowledge, is asking would someone be tricked or misled into believing you are the other guys. I think that is applicable here where someone drawing a talking mouse isn't infringing as long as it cannot be mistaken for Mickey Mouse, which again would be the fault of the person inducing the creation and not the tool that allowed it to happen.

Where does "inspired by" / derived from the encoded knowledge turn into outright exploitation of copyrighted work? There's certainly _a_ line but I find it difficult to define it at it being encoded into knowledge of it existing.

Everything that he sees has mysterious flaws that never happen.

I agree in principle, but that they can in the first place, especially when it accidentally happens, and at such massive scales more importantly, is the issue methinks.

And no one's talking about abolishing the AIs here, we're just talking about wanting M$/OAI to do their due diligence and get access to their training materials fairly. NYT wouldn't have sued if M$/OAI had approached them and struck a deal of some sort with them, but that's not what they did. They took in whatever data they could, from wherever they could baring no mind at all to where the data came from and what was being done with it.

There's a reason Getty images managed to strike a deal with Dall-E and why many of the image generation models now solely rely on data that is verifiably free of copyright (or where deals have been made in the case of Getty images). It's easier to see in pictures when a blatant copy is made (like watermarks) so it's obvious why Dall-E was the first to encounter this hurdle, but this was inevitable even for plain text that ChatGPT returns.

These things are indeed "a program on a machine and doesn’t have rights", but what I find scary is that rights aren't part of the rules of the universe, they're merely laws, created and enforced (to the extent that they are at all) by humans.