Comment by lakySK

If you treat LLMs as know-it-alls and just copy the output, then your expectations of the current generation of LLMs are too high.

That doesn’t mean they’re not useful though.