As AI adoption accelerates across human society, the problem of aligning AI models with human preferences remains a grand challenge. Currently, the AI alignment field is deeply divided between behavioral and representational approaches, resulting in narrowly a... ...