I've seen some that will annotate text but nothing that will fade AI generated text out to the background color depending on the level of confidence that a text was human written or AI written. Does anybody have something like this or have it in the works? If so is it local or does it require some service?
If that's the case then they have little use. We'll be reading AI created text from now on and we won't even know it. I'm not sure how I feel about that.
I wonder what % of wikipedia articles do have substantial portions of AI-generated text by now (and is it really a problem, providing the information is accurate - in fact in most cases I can't even imagine caring too much either way if something's written by a bot, but I do care if it's hallucinated BS).
I’m reminded of the (apocryphal?) story, about the Ranger at Yellowstone National Park who was asked why they didn’t just make all the bins bear-proof. His answer:
“There is a surprisingly large overlap between the smartest bears and the least smart humans.”
From everything I’ve read, outside those models that have plagiarism seeds that their own tools can detect, there is a surprisingly large overlap between the most humanlike AI and the most robotlike human writing.
