635
AI labs buy, scan, shred millions of rare books (www.news.com.au)
submitted 2 months ago* by to c/technology@lemmy.world
179 comments fedilink hide all child comments
 

Artificial intelligence labs are in a new arms race to buy up millions of rare books, slicing them open, scanning the pages and pulping the remains — sparking concerns that the last remaining copies of out-of-print texts are being destroyed on an industrial scale.

ISBNdb notes that “print books from the pre-LLM era are structurally guaranteed to be free of this contamination”.

“Millions of the most valuable books have never been digitised. They exist only in physical form, scattered across library shelves, used bookstores, and out-of-print catalogues. We get them to you at scale.”

you are viewing a single comment's thread
view the rest of the comments
[–] 9 points 2 months ago (2 children)

The bookseller explained that in April, he suddenly went from selling around 20 books per week to hundreds — including obscure and out-of-print works.

“It benefits me financially as well as by clearing out old inventory that is otherwise unlikely to sell,” he said.

I hate the AI slop as much as the next guy, but these are books nobody wanted. They were going to rot on the shelf until being thrown out. Just being rare doesn't mean it's particularly valuable.

Hopefully they are retaining the digitized versions for release at some point in the future. I'd imagine they are, since they'd still need the raw data to train future models.

  • permalink
  • fedilink
  • source
  • hideshow 2 child comments
    •
  • [–] 25 points 2 months ago (1 child)

    You can't determine what knowledge will be useful or valuable to someone in the future. Archaeologists have poured over receipts and shopping lists from ancient Egypt that were thrown away. It's why the Internet Archive backs up all the random webpages they do. There's no justification for mass destruction of books, especially since it's being done to train broadly useless LLMs.

    Of course they won't be releasing these books later on. These companies want to hoard this text data so their competitors can't access it. It would also be illegally distributing copyrighted work. Aside from destructive scanning being faster, the fact that a book is destroyed makes the digital copy "transformative" and therefore legal if it isn't further copied or distributed. A judge ruled on this recently. I don't mean to advocate for copyright enforcement, but that's the reality here.

    I think China has a national repository of AI training data that companies can access if they contribute to it as well. Doing the same thing in the West would at least blunt some of the harm the AI craze is causing, but that would have made too much sense.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 1 child comment
    •