AI companies are reportedly shredding millions of books after using them to train AI models — tech giants outsource to middlemen to secretly buy up books for tr

AI companies are reportedly shredding millions of books after using them to train AI models — tech giants outsource to middlemen to secretly buy up books for tr

SkyBill40 I don't understand why they'd destroy the books. :/ This situation reminds me of a t-shirt I saw recently: "Make Orwell fiction again." Reply

salgado18 SkyBill40 said: I don't understand why they'd destroy the books. :/ This situation reminds me of a t-shirt I saw recently: "Make Orwell fiction again." Inventory space for unused assets. That way, they don't even need a room to store them, and it gets cheaper. Always money. Well, sellers could create a different type of business: lend the book, then do a reverse-logistics to receive it back. Charge more for it, as it seems they don't care about costs. Book sellers keep inventories while turning a profit, preserve hard to replace books, AI companies don't look so evil, and everybody wins. I guess it's too late for that, but still, would solve many problems. Reply

drea.drechsler warezme said: Do we really want Stephen King as the basis for training AI without developing context into the AI? Actually: yes. AI's have to know what "popular culture" is to comment on it at all. What is REALLY funny is ask an AI about most any a Stephen King novel and it will agree it's pure trash, even if entertaining, or great literature. It all depends on how you ask the question. Reply

drea.drechsler SkyBill40 said: I don't understand why they'd destroy the books. :/ I see nothing wrong with this. You're allowed to buy a book and keep it for reference without any violation of copyright laws. What you're not allowed to do is make copies except for personal use. An AI does the same thing, but it has to have the 'book' digitized in a database…that's what its 'memory' is after all… in order to reference it and make fair-use quotes for servicing 'clients'. That digitized copy is probably a copyright violation that will get them in trouble. What I don't know is whether or not destroying the original copy of copyrighted material they legally purchased relieves them of the copyright violation for making digital copies (I am not a lawyer). Especially to use as AI's do. Also, look at the recent Anthropic settlement so see what's at play here. There were at least two parts: one is they got pirated material (never good) but the part that got the 1.5B dollars is they kept copies (digitized in the AI's memory/database) and used them. But at any rate it's no different from me reading the latest NY Times best seller on summer vaca then ripping pages out to start a fire in the cabin stove this winter. My library will not take donated books, they'd just toss them in a dumpster; they only want books they obtain and are in line with their collection management policies. Reply

bigdragon Humans write books on computers. Publishers buy dead trees to print their books on. AI companies buy up books to scan into their computer models. Then AI companies destroy the books made out of dead trees… Is there a way to do this while skipping the whole dead tree part? warezme said: Do we really want Stephen King as the basis for training AI without developing context into the AI? At least they're using the books and not the movies. I would hate to see what a Stephen King-inspired meatball recipe looks like! Reply

Key considerations

  • Investor positioning can change fast
  • Volatility remains possible near catalysts
  • Macro rates and liquidity can dominate flows

Reference reading

More on this site

Informational only. No financial advice. Do your own research.

Leave a Comment