Also, why are people acting like Anthropic digitizing and destroying a rare book is what makes it inaccessible to the public?
If they were simply buying the books and doing nothing with them, the public wouldn’t be able to access their copies anyway.
Because destroying the book makes it possibly inaccessible permanently because we have no idea how long anthropic plans on storing the digital copy, if at all. A lot of people assume they would for future training, but you don't know that.
If they at least didn't destroy the book, someone could purchase it when anthropic eventually goes belly up. Hopefully someone will at least be able to purchase their digital scan and hopefully the scan is of decent quality and clearly indicates the provenance of the text.
> we have no idea how long anthropic plans on storing the digital copy, if at all. A lot of people assume they would for future training, but you don't know that.
Books are considered super high quality training data. Anthropic has no reason to get rid of this data that 1. They’ve spent a ton of money on and 2. Will remain useful indefinitely for training LLMs.
> If they at least didn't destroy the book, someone could purchase it when anthropic eventually goes belly up.
The same applies for a digital scan? The information isn’t any more likely to be lost.