This is exactly why many of us have a problem with AI use. People drop large generated output and then say "please help review it". I have no idea how much effort you put into reviewing the content or how much knowledge you have of the domain. And depending on that context, the "please help review" can mean either "I'm fairly confident about the content, but need someone to proof-read it" or "I don't really know if any of this is correct and want someone else to do all the work". You gave us no context on this.
I also think AI is in fact not very good for learning in general, because if you lack knowledge in a domain... you have no idea how correct the output is. It's better to use it as a fuzzy search and ask it to assemble a list of links and pointers to *first-hand* resources... unless you have an easy way to verify the claims in the output, which seems rather difficult given the abstract nature of the proposed content. On Sat, 2026-08-08 at 11:19 +0000, Sasha Bogicevic via ghc-devs wrote: > Hey all, > I wanted to share ghc handbook I made using AI > https://devnull-org.github.io/ghc-handbook/ > > I know people have opposite views when it comes to AI usage but I > believe using AI to learn things is really poweful. This page might > help interested haskell programmers to explore ghc internals. Not > sure how useful it is for people already knowledgable about ghc but > at least the ghc notes are easy to access. > Anyway, hope this attracts new people to do some work on the > compiler, please let me know what you think or if anybody is > interested in reviewing the content. This is just the first draft so > future versions should be better. > > Thanks, Sasha > > > PGP 6023d9e35e208decad5101f0be284d15ea8108a2 > > Sent from Proton Mail for iOS. _______________________________________________ ghc-devs mailing list -- [email protected] To unsubscribe send an email to [email protected]
