boscolo.eth
@boscolo.eth
Curious, how is an AI any different than a human? Humans read books and train their mind to answer questions, just like an AI.🤔 https://techcrunch.com/2025/06/24/a-federal-judge-sides-with-anthropic-in-lawsuit-over-training-ai-on-books-without-authors-permission/
1 reply
0 recast
1 reaction
depatchedmode
@depatchedmode
And I repeat: now that we have a property rights substrate for media, why bother with traditional IP approaches anymore? Let’s focus on effective application of property rights. Which obviously has subjective bits that need to be litigated, but seemingly with more historical precedence and fewer moving parts. Anthropic is guilty of stealing books, not ideas.
1 reply
0 recast
0 reaction
boscolo.eth
@boscolo.eth
can you draw a line (figuratively) around the copy? No The words appeared in the pages of the book in the most probabilistic sense possible based on collective human knowledge. Humans are all just thieves remixing the same stuff, with the occasional new seed. Music is the best example of this.
1 reply
0 recast
1 reaction
depatchedmode
@depatchedmode
Yeah, but in this case they use and keep copies of books they didn’t pay for. Which is actually illegal. Using them to train the model is, imo, fine. But they need to buy the copy, or somehow acquire the rights.
1 reply
0 recast
1 reaction
boscolo.eth
@boscolo.eth
I foresee training data centers colocated near libraries...😬
1 reply
0 recast
1 reaction
depatchedmode
@depatchedmode
I think that is actually fine and maybe even desirable, depending on the ownership structure of the resulting output. Like, I don’t think an LLM owned by a corporation is allowed to get a library card. But a municipally owned post-trained model would!
1 reply
0 recast
0 reaction
boscolo.eth
@boscolo.eth
i was being kind of snarky, but your response made me think a university is in a prime position to do this training, since it IS in possession of much of this source material
0 reply
0 recast
1 reaction
