The ability to explain the reasoning behind the output of a machine learning system is limited, as the system may produce an output that persuades without explanation. Machine learning may achieve its purpose without providing humans with the specific reasoning used to arrive at a decision.