When wishful thinking worksΩ
Safely and usefully spectating on AIs optimizing over toy worldsΩ
Computational efficiency reasons not to model VNM-rational preference relations with utility functions
A comment on the IDA-AlphaGoZero metaphor; capabilities versus alignmentΩ
[Link]Logical uncertainty and mathematical uncertaintyΩ
Logical uncertainty and Mathematical uncertainty
More on the Linear Utility Hypothesis and the Leverage Prior
Value learning subproblem: learning goals of simple agentsΩ
Against the Linear Utility Hypothesis and the Leverage Penalty
Being legible to other agents by committing to using weaker reasoning systemsΩ