[Link post] Michael Nielsen's "Notes on Existential Risk from Artificial Superintelligence" — LessWrong