AGI safety from first principles

Embedded Agency

2022 MIRI Alignment Discussion

2021 MIRI Conversations

Iterated Amplification

Value Learning

Risks from Learned Optimization

Cartesian Frames

Some comments on the CAIS paradigm

Experiments in instrumental convergence

Hypothesis Subspace

"Why Not Just..."

Law-Following AI

The Shard Theory of Human Values

AGI-assisted Alignment

Selection Theorems: Modularity

Breaking Down Goal-Directed Behaviour

Basic Foundations for Agent Models

Pragmatic AI Safety

AI Races and Macrostrategy