It is similarly unsulting to read an ungrammatical blog post full of misspelings. So I do not subscribe to the part of your argument "No, don't use it to fix your grammar". Using AI to fix your grammar, if done right, is the part of the learning process.
A critical piece of this is to ensure it is just fixing the grammar and not rewriting it in its own AI voice is key. This is why I think tools like grammarly or similar still have a useful edge over just directly using an LLM as the UX let's you pick and choose which suggestions to adopt. And they also provide context on why they are making a given suggestion. It still often kills your "personal voice", so you need to be judicious with its use.
Much less insulting than AI slop.
I can imagine it’s hard to see the nuance if you’re ESL but it’s there.