[not_kurt_godel]

My experience is it's been unbelievably fun until I actually try to run what it writes and/or add some non-trivial functionality. At that point it becomes unbelievably un-fun and frustrating as the LLM insists on producing code that doesn't give the outputs it says it does.