I don’t see how the machine that generates code that must be assumed to be broken, and statements about said code that must be assumed to be wrong, until both have been manually verified, because the “hallucination” issue is inherent to the design of LLMs and cannot be solved, is supposed to make my life easier. Especially since a well known problem in computer science is that writing code is more fun than reading it.

quoted for emphasis. you have more patience than I do in telling sycophants why their take is idiotic.