x
Chain-of-Thought Alignment — LessWrong