Non-coercive motivation for alignment research? — LessWrong