Preference Learning for AI Alignment: a Causal Perspective | Read Paper on Bytez