Done this as well. But after the initial "wow!" moment, the "make it better" part became a "actually I don't like how you wrote it, it doesn't sound like me".
There is a thin line between enhancing and taking over, and IMO the current LLMs cross it most of the time.
Your mistake was having the AI rewrite at all, don't do that, that is exactly the problem with them - that is them thinking for you. Ask the AI how well you wrote it, make the AI a writing professor, make the AI adopt a Socratic attitude that does not do anything but draws you to make the connection yourself.
There is a thin line between enhancing and taking over, and IMO the current LLMs cross it most of the time.