Cremer-Schutter2021 Article HowToUseAlgorithmicDecision-ma
This peer-reviewed article (De Cremer & De Schutter) examines how algorithmic decision-making is used in recruitment and the risks it poses to inclusiveness. It explains that algorithms can reproduce historical biases (e.g., Amazon hiring tool) and offers practical solutions for data scientists—primarily rebalancing training data per subgroup and adjusting evaluation via cost–benefit matrices—to reduce disparate error rates across subpopulations. The paper also argues for collaboration between data scientists and domain experts and organizational policies/training to embed inclusiveness.
Questions this document answers
10.1007/s43681-021-00073-0
Received: 16 February 2021; Accepted: 13 June 2021; Published online: 22 June 2021
David De Cremer (NUS Business School, National University of Singapore) and Leander De Schutter (Rotterdam School of Management, Erasmus University, Rotterdam, Netherlands)