cross-posted from: https://kbin.earth/m/piracy@lemmy.dbzer0.com/t/3120649
As AI tech companies increasingly buy and destroy books to feed to their AI models, Anna’s Archive is calling for volunteers to help preserve them for the public record.
cross-posted from: https://kbin.earth/m/piracy@lemmy.dbzer0.com/t/3120649
As AI tech companies increasingly buy and destroy books to feed to their AI models, Anna’s Archive is calling for volunteers to help preserve them for the public record.
I agree about not destroying books, but Anna’s archive is ok with providing all of their info to llms
https://web.archive.org/web/20260305153357/https://annas-archive.gd/blog/llms-txt.html
Not like it’s their books to begin with. It’s all available as torrents
Eh, perfect is the enemy of good. I’ll take the wins we can get. It’s not like having to buy the books is gonna stop the biggest offenders from burning investor money (along with the books).
The defeat of LLMs isn’t gonna come on that front. Let the cultists believe that more words will somehow become more than words. Their silicon gods run on power and water. That’s where they’re most vulnerable.
It’s not necessarily trying to stop them burning the books, just stop their ability to monopolise the knowledge, that’s their aim, lock the knowledge behind a subscription, Anna’s archive will always be available for all, so why pay for their subscription if its available for free.
I was aiming for the poetic parallel between burning money and destroying books, but also, isn’t the difference between destroying the primary source and preventing access to it mostly academic? Either way, as you say, the AI grifters are looking to control that information. Once they do, the original text and knowledge might as well be lost, since we couldn’t verify anything without the original to compare to. Whether through deliberate human manipulation or random word soup mutation, they can’t ever be trusted to deliver a genuine, faithful reproduction of the contents.
And the conclusion is the same: Better to have it freely available for everyone (including the grifters) than the grifters only.
They’ll come for it at some point.
I am all up for starving machine learning of new knowledge but I view this as the least friction path. Otherwise, if the machines try to brute force their entry, what harm can kt cause?