Stuart_Armstrong(Stuart Armstrong)

Sleep­ing Beauty gets coun­ter­fac­tu­ally mugged

Avoid­ing dooms­day: a “proof” of the self-in­di­ca­tion assumption

An­thropic rea­son­ing and cor­re­lated de­ci­sion making

Cor­re­lated de­ci­sion mak­ing: a com­plete theory

Ex­treme risks: when not to use ex­pected utility

Ex­pected util­ity with­out the in­de­pen­dence axiom

Post re­tracted: If you fol­low ex­pected util­ity, ex­pect to be money-pumped

Money pump­ing: the ax­io­matic approach

All hail the Lis­bon Treaty! Or is that “hate”? Or just “huh”?

Con­se­quences of ar­bi­trage: ex­pected cash

In con­clu­sion: in the land be­yond money pumps lie ex­treme events

Ad­vice for AI makers

The AI in a box boxes you

The Black­mail Equation

Black­mail, Nukes and the Pri­soner’s Dilemma

Omega’s sub­con­tract­ing to Alpha

SIA won’t doom you

The Last Number

AI in­differ­ence through util­ity manipulation

Let’s split the cake, length­wise, up­wise and slantwise