Jan SchieferIt will be interesting to see which target languages are more suitable for LLM-based code generation. Intuitively, I would think that the more strict and typesafe languages, such as Rust, Ocaml, F#, would fare better. If the compiler can stop me from doing something stupid, surely this must work for AI-generated code also?
LogicalErzor@janschiefer i think ai will have a massive impact in language design and reveal much needed niceties in programming:
eg.
- concise error messages
- database/git undo
- sensible flags to commands
- etc.
these would be nice to have as humans, but prob would make it necessary for ai as it would be much easier to audit ai and undo errors
@Logical_Error Indeed, but there is a bit of a chicken and egg problem here: Unless it shows up prominently in a training data set, it won't matter, because the LLM won't know what to do with it.
Vaguely related, has anybody tried to use Claude Code to write something in INTERCAL, or Befunge-93? That might be a fun experiment.