I'm not saying you're entirely wrong, but LLMs can actually be a way to create even better things than those better things. I've seen it in action.
How do you judge "better" if you don't understand the LLM output?
How do you judge "better" if you don't understand the LLM output?