So you think training should be legal, but actually using the trained model to generate copyrighted text should be illegal?
Edit: As a secondary point, it's not like Meta bought all those books. They downloaded unlicensed copies.
Edit: As a secondary point, it's not like Meta bought all those books. They downloaded unlicensed copies.
You can use books without buying them. e.g. libraries, or borrowing books from anyone even if they're not a "library".
Copyright is not about internal use, it's about copying and distribution.
This is not about what anyone thinks should be legal, but rather what is legal under the current law. The law was not designed for the digital era where "use" could be something other than a person consuming the content, and this has not been meaningfully addressed, therefore other uses are legal by default because that's how the law works.
How did they scan them in then?
By your logic, distributing the digits of pi could be construed as copyright infringement otherwise.
of course you can - pi contains all known combinations of digits.
On the other hand, the "address" of a text within a memorizing large language model would just be the prompt "give me the text of X".
The ease with which these "retrieval" operations can be done is irrelevant.