Hacker News Reader

If they have no value why are they being acquired and scanned?

A flagship LLM today is trained on tens of trillions of tokens, the equivalent of hundreds of millions of 100k word books. No human has that kind of appetite.

As an aside, the entire Google Books corpus is generally estimated at tens of millions of books.

Any original text has value for the purpose of AI training. The argument is that they have no other value.

Exactly. It’s the funniest thing. These books are only being bought by AI companies. It seems pretty clear they only have value to AI companies. Otherwise all these people bemoaning the loss of rare books would be… buying them.

“How dare these companies buy rare and valuable books that nobody else values enough to buy” is a self-canceling argument.

Reply on news.ycombinator.com