If an AI came to have a self, at what point should we start treating it as a moral being?