Sadly, it is common enough these days. There are reports of authors that set up a stand to physically sell their books, and members of the public ask "is this AI-written? Who made the cover? Is this cover AI?"

> My impression is that when it comes to reading text, nobody cares as long as the final product is good.

Maybe so, but it only takes a vocal minority to ruin things for an author. And that minority tends to use anything, even a non-overtly-critical mention of AI in social media.

The consensus right now is that using LLMs for fixing grammar and typos is acceptable. I personally use them for word completion (specially the devil incarnate which are the prepositions on/in), but tend to discard suggestions that improve flow, sentence structure and readability, because those increase the odds of triggering "AI detectors". In fact, I've found a renewed taste for unconventional sentence structure and unconventional punctuation; things that three years ago, before the LLM boom, I really didn't care for.