An outsider can throw test data points at the model and see how it behaves. For example, whether changing name, race, sex, or variables correlating with such cr

An outsider can throw test data points at the model and see how it behaves. For example, whether changing name, race, sex, or variables correlating with such critical variables, alter a decision. Pointing out these flaws could then lead to better systems, or not using machine learning on some case where input data is unsuitable.