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.