Trust Levels

Are people trusting agents with decision-making yet, or still keeping a human in the loop for everything important?