For me, the value in posting unedited output is that the human is not messing with what the A.I. said - especially not making it appear that the A.I. output supported a view that, unedited, the output clearly didn't.Sage wrote:...
I know this rule wasn’t made because of me. It was made because A.I., when used carelessly or impersonally, can flood conversations with volume instead of value. MG 2.0’s approach — plugging in prompts, posting unedited output, and not consistently engaging with others — shaped the perception of what A.I. is and how it behaves here.
...
Unfortunately, I think, it's not feasible to cite A.I. output. And because (if I'm not mistaken) AIs are constantly absorbing new information (information gain), it's not possible to be sure that the same human asking the same A.I. the same question twice will get an identical answer the second time. Just like asking a person the same question on two separate occasions. Unless the answer has been prepared, and is regurgitated verbatim, you're likely to get answers that vary - sometimes only in fine details, and sometimes to the extent of being radically differnt. Just look, for example, at how physicists would answer questions about protons and neutrons before and after the exposition of quantum chromodynamics. (waiting for PG to correct me on this
