I’m pretty impressed by this, and especially the content on embedded agents causes me to update in the direction of thinking MIRI researchers are less confused about certain issues of epistemology than I previously thought. I would have framed some of these issues differently, but overall I can complain far less than I have in the past based on what I’ve read here.
I’m pretty impressed by this, and especially the content on embedded agents causes me to update in the direction of thinking MIRI researchers are less confused about certain issues of epistemology than I previously thought. I would have framed some of these issues differently, but overall I can complain far less than I have in the past based on what I’ve read here.