Can a Robot Know When It’s Wrong?