As Graehl and wedrifid observed, Eliezer responded as if the original statement were an absolute. He applied deductive reasoning and found a reductio ad absurdum. But if, instead of an absolute, you see multifoliaterose’s characterization as a reference class: “People who believe themselves to be one of the few most important in the world without having already done something visible and obvious to dramatically change it,” it can lower the probability that Eliezer is, in fact, that important by a large likelihood ratio.
Whether this likelihood ratio is large enough to overcome the evidence on AI-related existential risk and the paucity of serious effort dedicated to combating it is an open question.
As Graehl and wedrifid observed, Eliezer responded as if the original statement were an absolute. He applied deductive reasoning and found a reductio ad absurdum. But if, instead of an absolute, you see multifoliaterose’s characterization as a reference class: “People who believe themselves to be one of the few most important in the world without having already done something visible and obvious to dramatically change it,” it can lower the probability that Eliezer is, in fact, that important by a large likelihood ratio.
Whether this likelihood ratio is large enough to overcome the evidence on AI-related existential risk and the paucity of serious effort dedicated to combating it is an open question.