https://norbu-ai.org/ is an LLM with expertise focused specifically on the Pali Canon. It’s supposed to have narrow expertise and be used for a single purpose.
Norbu is useful for searching through the canon and say, “Where does the Buddha talk about alcohol?” “Where does the idea of the seven factors of awakening come from?”
Norbu allows you to get from concepts you’ve heard about back to primary sources. It does have some advantages over just searching suttacentral.net or accesstoinsight.org: you don’t have to know the keywords you’re looking for, you can look for things you’re only starting to understand when you don’t know the jargon.
The works of Marx and Engels are about the same size as the Pali Canon. Imagine a narrowly-expert LLM, a fine-tuned open-source model, and baby leftists asking it things like –
- Where is the tendency of the rate of profit to fall discussed?
- What was Marx’s understanding of international trade? (and it wouldn’t just tell you what it was but what chapter and verse it’s citing)
- Did Marx ever consider there wouldn’t be gold-backed money?
and the LLM gives answers grounded in the primary sources.
What do you think of my idea?
i like it. that’d be actually great for referencing works, and finding stuff by asking it questions.
i don’t think it’d be that hard if you meet the steep hardware requirements for doing it.
one could probably do something like that if enough people are interested.
I’d like it. Making the literature more accessable is a good thing imo. Important to me would be the trustworthiness you mentioned. This includes showing the specific source material but even more important: no hallucinations. Just tell me if there’s nothing there, don’t make shit up.
“no hallucinations” isn’t a thing within the LLM paradigm
Norbu always says “I may be mistaken, please verify”
it’s definitely doable, you can install a local model and train it on marxist literature, you probably just need like 64gb of ram lol. maybe 32 will do tho. but yeah, it’s not a cheap project i guess.
