From the LLM's "point of view" it is going to do what characters in the training data were most likely to do. I have a lot of issues with the framing of it hav

From the LLM's "point of view" it is going to do what characters in the training data were most likely to do. I have a lot of issues with the framing of it having a "point of view" at all. It is not consciously doing anything.