This could certainly take place if machine learning is used inappropriately. Machine learning versions are built to make "predictions" which have been only legitimate if we suppose the future will resemble the earlier. If they are trained on data that includes the outcomes of racist choices up to now, machine learning designs will have to forecast