The current approach leaves it frustratingly judgmental and prone to lecturing the user about ethics from a very particular point of view (yes, I am aware the system has no conscious intention, but the abstractions work from the user's point of view). In that regard they are simulating a type of person quite well.