Re:How to Get ChatGPT to Stop Apologizing?
by techdragon on Tuesday August 01, 2023 @01:58AM
(#36952828)
I think about it from a “what would a hypothetical random person say”… it’s not intrinsically holding a view or set of facts, instead it’s holding a weighted average sum of what people have said and written, including all the conflicting viewpoints inherent in that.
I suspect the question of what an LLM _really_ thinks has an answer that looks more like a probability distribution than typical human opinions do. (Actually, even humans give different answers on polls, depending on how the question is asked.)