LLMs currently write code. Some of that code is probably better than some humans code. In fact, LLM-generated code is useable, and as I'll demonstrate, rather scary good. But the code that LLMs write - is it good code? Is it safe, secure, well-commented, etc...? These are important questions, I'll grant you, but I think they are the wrong questions. I think the question should be: Where does the human fit into the LLMDLC (LLM Development Lifecycle)? (Cuze LLMDLC is coming, if not here already) Do we simply feed an LLM a set of requiremnts and turn coding over to it? Do we trust it to generate the right code, secure code, safe code, correct code? Is the human mearly become a code reviewer?