[WolfeReader]

Honestly, this. The mainstream coding culture has spent decades dealing with shoehorning stateful OOP into distributed and multithreaded contexts. And now we have huge piles of code, getters and setters and callbacks and hooks and annotation processers and factories and dependency injection all pasted on top of the hottest coding paradigm of the 90's. It's too much to manage, and now we feel like we need AI to understand it all for us.

Meanwhile, nobody is claiming vast productivity gains using AI for Haskell or Lisp or Elixir.

[lukev]

I mean, I find that LLMs are quite good with Lisp (Clojure) and I really like the abstraction levels that it provides. Pure functions and immutable data mean great boundary points and strong guarantees to reason about my programs, even if a large chunk of the boring parts are auto-coded.

I think there's lots of people like me, it's just that doing real dev work is orthogonal (possibly even opposed) to participating in the AI hype cycle.