logoalt Hacker News

bonesss • yesterday at 8:52 PM • 1 reply • view on HN

Some large percentage of the dialogue around LLM prompting boils down to arguments we should be speaking to them in Haskell.


Replies

black_knight • yesterday at 10:40 PM

They understand Haskell (and even Agda) remarkably well. One must just get their context out the degenerate state where they use strings, ints and bools for everything, and then they will happily write truly robust pieces of code!

They can even write good property based tests, with impressive Arbitrary instances for everything!