Because there's a mental overhead when you're not writing the code that is arguably worse than when you are writing the code. No one is talking about this enough IMO but that's why everyone is so exhausted when using LLMs and end up just pulling the slot machine until it works without actually reading it.

Reading code sucks, it always has. The flow state we all crave is when the code is in our working memory as an understood construct and we're just translating our mental model to a programming language. You don't get that with LLMs. It devolves into prorgamming minutae equivalent to "a little to the left" but with the added complexity that "left" is hundreds of lines of code.