[rootnod3]

I must be using it wrong then or using the wrong languages. All I have seen it produce so far was mediocre at best and painfully wrong 2-3 prompts in.

Not even starting with how it just “fixes” a hug by introducing a wholly new one and then re-introducing the old one when pointing it out.

[aurareturn]

You most likely are.

Or maybe the LLM just hasn't been trained enough on the language you're using.

[EForEndeavour]

Obligatory but important: which model, application, and programming language was this in?

[yomismoaqui]

Let me guess... GPT 3.5, ChatGPT and Brainfuck?

[rootnod3]

Claude 4.5, all variants, Rust, C, Common Lisp and Haskell.