+- The Voynich Ninja (https://www.voynich.ninja)
+-- Forum: Voynich Research (https://www.voynich.ninja/forum-27.html)
+--- Forum: Voynich Talk (https://www.voynich.ninja/forum-6.html)
+--- Thread: AI Companies destroying rare books to train their AIs (/thread-5980.html)
RE: AI Companies destroying rare books to train their AIs - Rafal - 05-08-2026
For me the problem is not that they are destroying books.
First they destroy books that they bought and paid for. It is their right to do whatever with them.
Then, they are not destroying rare manuscripts but single copies of modern books that were printed in large quantities.
Personally I don't treat books, unlike some people, as something holy. You don't commit a mortal sin if you destroy or trash a book.
The problem is that they train AI with these books without paying a single penny to their authors. Here things became immoral.
RE: AI Companies destroying rare books to train their AIs - Koen G - 05-08-2026
100% agree, Rafal. The vast majority of books aren't rare at all. When books are actually rate, they'll be in collections. The rest, as long as it's not in a store, library or personal bookshelf, is just paper taking up space. We're not talking about unique early prints or manuscripts here.
The problem is the practice. Steal everything humanity has ever produced and turn it into a questionable product.
RE: AI Companies destroying rare books to train their AIs - Aga Tentakulus - 05-08-2026
There was a report on it recently.
Here’s the link, whilst it’s still working.
You are not allowed to view links. Register or Login to view.
RE: AI Companies destroying rare books to train their AIs - bi3mw - 05-08-2026
I’d accept the argument “I can do whatever I want with a book I’ve bought” when it comes to individuals, but not when a corporation is making purchases on a massive scale. The problem isn’t that old books sometimes end up in the trash. The problem is the system behind it. Antiquarian bookstores function like a kind of circulating library. An old book is sold, read, and resold. Then, years later, it resurfaces and finds new readers. This keeps knowledge accessible, including niche knowledge, old editions, and out-of-print titles that can’t be found anywhere else. Now, AI companies are buying up these books by the pallet. In other words, they’re systematically purchasing entire inventories, scanning them for AI training, and then destroying them. This is an unacceptable “one-way street”; in the medium term, secondhand bookstores and individual customers will lose an entire segment of the book market, resulting in the loss not only of junk but also of cultural heritage. Furthermore, it’s clear that the value attributed to a book is always in the eye of the beholder. It isn’t always measured by the price one has to pay for it. So what one person sees as “just paper taking up space” is something valuable to another. However, only individuals can make this decision. That’s how the market in this field normally works.
RE: AI Companies destroying rare books to train their AIs - Koen G - 05-08-2026
If this went hand in hand with digitization efforts (for books that are not copyrighted anymore), at least it would help to spread the knowledge. Something like what Google's been doing. And there are book types that exist in excessive supply. Dated encyclopedias come to mind, but also many others. Anything that's been printed for mass market can take a hit. So for me it really depends on what these supposedly rare books are.
But the voracity and, I can only presume, lack of selection criteria on which books are ingested, is worrying indeed.
RE: AI Companies destroying rare books to train their AIs - bi3mw - 05-08-2026
According to a bookseller, the selection criteria are extremely broad:
Quote:The focus was specifically on non-fiction titles published after 1970 with ISBN numbers.
The dealer himself describes these books as worthless. The only way I can explain this is that he’s trying to justify his questionable sale in hindsight.
RE: AI Companies destroying rare books to train their AIs - Koen G - 05-08-2026
(05-08-2026, 09:18 PM)bi3mw Wrote: You are not allowed to view links. Register or Login to view.non-fiction titles published after 1970 with ISBN numbers.
I just feel like this kind of book would eventually end up being discarded anyway. I have turned in many myself to the second hand store, where they were drowning in that kind of stuff. Unless it's a first edition or something, in which case it will be in collectors' hands. Or maybe I am not imaginative enough to picture the kind of mass-print book published after 1970 that would suffer from having one or two of its copies destroyed.
This is not to say that the project is a noble cause, but "rare books" made it sound like they were burning through books of hours and incunabula.
RE: AI Companies destroying rare books to train their AIs - bi3mw - 05-08-2026
(05-08-2026, 09:24 PM)Koen G Wrote: You are not allowed to view links. Register or Login to view.Or maybe I am not imaginative enough to picture the kind of mass-print book published after 1970 that would suffer from having one or two of its copies destroyed.
Well, some reprints of old works (with an ISBN) might very well have been published after 1970. They would meet the selection criteria formulated by AI companies (even if they aren’t really usable). I think it would be better if such products didn’t end up in the shredder but remained available on the market.
RE: AI Companies destroying rare books to train their AIs - Pierre Dumont Himself - 05-08-2026
There is no evidence that the company only buys books with ISBNs, only that those are statistically more accessible.
Quote:As an internal memo from April 2024 shows, its goal was to scan and destroy "all the books in the world"You are not allowed to view links. Register or Login to view.
RE: AI Companies destroying rare books to train their AIs - Jorge_Stolfi - 05-08-2026
(05-08-2026, 10:38 PM)Pierre Dumont Himself Wrote: You are not allowed to view links. Register or Login to view.There is no evidence that the company only buys books with ISBNs, only that those are statistically more accessible.
Quote:As an internal memo from April 2024 shows, its goal was to scan and destroy "all the books in the world"You are not allowed to view links. Register or Login to view.
They naturally want to prevent their competitors from scanning the same books, if they can.
There should be a law... But unfortunately there is no more Law in the US, only unchecked Greed.
There was a saying "nothing is illegal if twenty businessmen decide to do it".
Now just one is enough.
All the best, hopefully, one day... --stolfi