Comment by zozbot234

> accidentally conjure a horde of demons.

If we have a certain perspective on demons as self-sustaining information processing loops—we’ll, yeah, demons abound.

And, yes, this hn post from BenBreen — is amazing. We’ve been in touch!

And if Ben is reading, we’ve made progress on our version controlled community—LLM book translation prototype. Of course, it’s much to early to share on hn. Or whatever, it’s great, take a peek: https://www.philosopherslibrary.com/

Upload books, get translations. We have a separate system where neolatin scholars can evaluate randomly selected paragraphs — so we can measure and report on the base rate of different qualities for each book as a whole.

I don’t know if it is technically important for advancing AI, but it might be.

80% of the Neo-Latin books in the library have not been translated. Most NeoLatin hasn’t been translated. And it is even worse with Sanskrit. So much material has not been digitized or translated.

Is it meaningful to try to get this humanist core into our language models? Maybe including only modern writing is a good bias, but I doubt it.

It will take at least 2 years to get all this stuff scanned, digitized, translated and published.

So it’s kind of urgent, given all the AI training taking place over the next two years.

I’m writing this in a rush, but I’m at the Houghton Library at Harvard and they just brought up 5 books from Marsilio Ficino, published 1497 and 1516.

(If you or anyone is into this kind of thing, my email is in my profile — it’s my favorite hobby)