x
Structural Alignment: Why AI Needs Domain-Governed Architecture (Not Just Better Training) — LessWrong