AI-generated text has much to commend it. It's a fast, cheap (or free) and reliable way of generating content. It reads quite naturally, and it's here to stay.
I'm all in favour of getting AI to make things easy for us, but nobody can deny that it has some big issues.
It's not infallible. It can make mistakes. I asked a well-known AI text generator to tell me who I am, and I was told I was an Australian actor and an Australian football coach. It's not unreasonable that it got that wrong: I'm far less famous than either of them, though it surely should have spotted the difference in spelling between my name and the actor (in editing terms, it introduced an error). It has introduced factual errors into other things I've asked it, too, including telling me the wrong date for the founding of an organisation. These are errors that a human editor will most likely pick up.
It is unresponsive to the audience. One of the first questions a human editor will ask is: who will read this? AI doesn't care who's reading it; a human editor does.
It is unresponsive to your needs. You might have a particular purpose for writing something. Perhaps you need
It doesn't let your voice shine. Importantly, if you're writing something, it needs to be said in a way that is yours.
It may well also be unethical or illegal for you to use. Students in particular will likely face severe penalties if found to have used AI for assessed work.
