The ultimate test would be application. Can it replace humans in all situations (or at least all intellectual tasks)?

GPT4 sets pretty strong conditions. Ethics in particular is tricky, because I doubt a self-consistent set of mores that most people would agree with even exists.