You must log in or register to comment.
I bet that analogy sounded smart in their head
What part of buying rare books and destroying them after feeding them to the model training makes the analogy invalid?
They’re not targeting rare books. According to the article they’re buying books that have an ISBN, which means they’re in a global database and probably more likely to be common. The article does say there’s uncertainty about whether the people doing this filter out rare books. (because they seem to buy even when a book is overpriced)
Do I like this? No. But describing it as burning the library of Alexandria is dramatic and misleading. We’re not losing knowledge forever. But a lot of physical books are being taken out of circulation for a dubious return of value.
This article I just saw elsewhere on lemmy does claim that the books are rare. But it seems to be misleading. In one point it talks about Dutch antiquarian book collectors receiving requests. But if you click through to the source story (Dutch) it says, “Usually the spam button was quickly pressed, because it was hardly serious to take. Antiquarians only deal in very old books: titles that stem from the time before ISBN codes existed at all.”
Again, I don’t like what these companies are doing, and wouldn’t be opposed to legislation against it. But it’s good to be factual and understand what is actually happening.
Having an ISBN number doesn’t guarantee there is a digital copy somewhere else or that a significant number of physical copies exist. But it makes it more likely.
Says the dude making the post, posting himself on linkedin
Say the AI companies themselves. They are openly buying physical books, prioritising rarer ones, tearing them apart to digitise them, and then burning the pages.
