Yes, LLMs are more like offline code generators that can't be reliably re-run. So the very first step of producing the code is "easy", but after that you have l

Yes, LLMs are more like offline code generators that can't be reliably re-run. So the very first step of producing the code is "easy", but after that you have lost that ease, and have to read and maintain the larger generated output.