That implies that "vulnerability/fallibility, social reputation/integration, slowness/interpretability, empathy" are not useful for decision making. Until we see intelligence comparable to human I don't think we can make that assumption.