An AI, even if it’s capable of manipulating the world beyond our understanding, will likely opt for non-chaotic plans over chaotic plans. A plan built on control of a web of people via emotional manipulation is still subject to the chaos of random things happening to those people, differences in personality, the full entropy of the human experience. What is predictable is an X-maximizer will seek plans that minimize chaos, thus it’s preference for those plans would not seem as weird to us.
An AI, even if it’s capable of manipulating the world beyond our understanding, will likely opt for non-chaotic plans over chaotic plans. A plan built on control of a web of people via emotional manipulation is still subject to the chaos of random things happening to those people, differences in personality, the full entropy of the human experience. What is predictable is an X-maximizer will seek plans that minimize chaos, thus it’s preference for those plans would not seem as weird to us.
I agree with this take, but do those plans exist, even in theory?