Unpacking the Ethical Value Alignment in Big Models