A Com­ment on Ex­pected Utility Theory

Math­e­mat­i­cal Sys­tem For Calibration

Hu­mans are not agents: short vs long term

New cir­cum­stances, new val­ues?

Op­ti­mi­sa­tion in ma­nipu­lat­ing hu­mans: en­g­ineered fa­nat­ics vs yes-men

Acausal trade: trade barriers

Agents that don’t be­come maximisers

Un­der­stand­ing the im­por­tant facts

Trans­la­tion “coun­ter­fac­tual”

In­differ­ence and com­pen­satory rewards

The ra­dioac­tive bur­rito and learn­ing from pos­i­tive examples

Cor­rigi­bil­ity thoughts III: ma­nipu­lat­ing ver­sus deceiving

Cor­rigi­bil­ity thoughts II: the robot operator

On­tol­ogy, lost pur­poses, and in­stru­men­tal goals

How to judge moral learn­ing failure

The best value in­differ­ence method (so far)

Re­ward/​value learn­ing for re­in­force­ment learning

An­nounc­ing AASAA—Ac­cel­er­at­ing AI Safety Adop­tion in Academia (and el­se­where)

Am I/​Was I a Cul­tist?