I personally always feed my blog posts into an LLM, but I only let it point out issues: I never apply the suggestions it gives me. What I mean is that if it gives me a paragraph or sentence to replace inside my article I won't copy paste his suggestion in the text, I will review the problematic section myself and, if I agree that it must be revised, rewrite it on my own. Ideally, this would happen without even reading the AI's suggested replacement.