Let's also not forget that humans are more than capable of creating tons of 'slop' far worse than any current SOTA model would create, and I would argue the majority of what people write is in this category.

Models are still improving, and if you can't tell whether one was behind what you're currently reading 100% of the time (which I'm sure you know you can't), the distinction no longer matters except in very specific cases, and you have to go to a lot more effort to uncover that distinction.