Eliezer: We have argued about evolutionary psychology for a long time. Maybe we need a disagreement case study.
“But human beings do find power over others sweet, ”
I would be careful to distinguish “power to accomplish goals”, which is a “basic AI drive”, and thus needs no evolutionary explanation in an intelligent system, or even in a system built by conditioning, and power specifically over others, which does. It seems compelling that both are desirable, but the latter doesn’t need to be for power to be somewhat corrupting in the manner discussed.
It doesn’t really seem likely that there’s much evolutionary basis for wanting power over MANY others though. You probably get that from reflective generalization of evolutionary preferences. This will lead to a wide distribution where presidents and tyrants, who have fanatically sought power, will find it VERY attractive, even when actually holding it prematurely ages them and leaves their lives in shambles. Kings, by contrast, will often casually give up power over large numbers of people, or possibly even not really notice that they have it, leaving rule to their advisers and officials so long as their basically rather cheaply fulfilled (by a nation) evolved goals (most significantly status in all likelihood) are met. Certainly Washington was an exception for a revolutionary, but it does look to me like the crown of the United States was not much of a temptation. What could kingship over a couple million rowdy backwoodsmen have given him that ownership of a thousand slaves couldn’t? Yet the latter was SO MUCH easier, safer, less demanding etc, and he was old and childless and left power to trustworthy comrades in arms.
A better case for generally corrupting influences from power comes from petty officials, who have not been selected to the degree to which rulers have, acting capriciously. This however can also be interpreted as resentment of rules or of rule based nominal authority over them, or alternatively as laziness, or to imitating others and filling a social role. Once again, subtle conditioning is also a possibility. Just ask any animal trainer how easily bored animals learn to play potentially annoying games with one another or with the humans. It falls out of a very general drive to learn/explore/experiment that practically any activity at the right level of difficulty or producing a highly salient response within the right range of complexity and predictability can be a reinforcer. If we, like crows, are evolved to test boundaries, to investigate what is safe, what our abilities allow, what we can get away with… if we are attracted to the right level of apparent danger, all this will make power corrupting in just the manner Eliezer discovered with the reporter without requiring any distinct evolved power drive.
Hmm. Does Atheism produce a corrupting feeling of power at first? It fits this model.
Its worth noting how very few people seek power on any significant scale, and how culturally and cognitively ill equipped to hold it most people are.
The human specific part of the genome is small, and omits simple and useful survival details such as the proper techniques for crocodile wrestling. Baboons and toddlers don’t even understand hiding! They have to LEARN to HIDE! I need a evolutionary explanations to make specific predictions that my other knowledge of psychology omits before predicting mutational support for a behavior.
Eliezer: We have argued about evolutionary psychology for a long time. Maybe we need a disagreement case study.
“But human beings do find power over others sweet, ”
I would be careful to distinguish “power to accomplish goals”, which is a “basic AI drive”, and thus needs no evolutionary explanation in an intelligent system, or even in a system built by conditioning, and power specifically over others, which does. It seems compelling that both are desirable, but the latter doesn’t need to be for power to be somewhat corrupting in the manner discussed.
It doesn’t really seem likely that there’s much evolutionary basis for wanting power over MANY others though. You probably get that from reflective generalization of evolutionary preferences. This will lead to a wide distribution where presidents and tyrants, who have fanatically sought power, will find it VERY attractive, even when actually holding it prematurely ages them and leaves their lives in shambles. Kings, by contrast, will often casually give up power over large numbers of people, or possibly even not really notice that they have it, leaving rule to their advisers and officials so long as their basically rather cheaply fulfilled (by a nation) evolved goals (most significantly status in all likelihood) are met. Certainly Washington was an exception for a revolutionary, but it does look to me like the crown of the United States was not much of a temptation. What could kingship over a couple million rowdy backwoodsmen have given him that ownership of a thousand slaves couldn’t? Yet the latter was SO MUCH easier, safer, less demanding etc, and he was old and childless and left power to trustworthy comrades in arms.
A better case for generally corrupting influences from power comes from petty officials, who have not been selected to the degree to which rulers have, acting capriciously. This however can also be interpreted as resentment of rules or of rule based nominal authority over them, or alternatively as laziness, or to imitating others and filling a social role. Once again, subtle conditioning is also a possibility. Just ask any animal trainer how easily bored animals learn to play potentially annoying games with one another or with the humans. It falls out of a very general drive to learn/explore/experiment that practically any activity at the right level of difficulty or producing a highly salient response within the right range of complexity and predictability can be a reinforcer. If we, like crows, are evolved to test boundaries, to investigate what is safe, what our abilities allow, what we can get away with… if we are attracted to the right level of apparent danger, all this will make power corrupting in just the manner Eliezer discovered with the reporter without requiring any distinct evolved power drive.
Hmm. Does Atheism produce a corrupting feeling of power at first? It fits this model.
Its worth noting how very few people seek power on any significant scale, and how culturally and cognitively ill equipped to hold it most people are.
The human specific part of the genome is small, and omits simple and useful survival details such as the proper techniques for crocodile wrestling. Baboons and toddlers don’t even understand hiding! They have to LEARN to HIDE! I need a evolutionary explanations to make specific predictions that my other knowledge of psychology omits before predicting mutational support for a behavior.