Speaker: Florian Keppeler, Assistant Professor, Department of Political Science, Aarhus University

Abstract: Applications based on artificial intelligence (AI applications) play an increasing role in the public sector decision making. They have the potential to fundamentally affect public personnel management, for example by affecting the selection of job candidates. While some research argues in favor of deploying AI applications to enhance efficiency and impersonality of recruitment decisions, other studies point to challenges of algorithmic decision-making and risks of increased discrimination. This study analyzes these positions based on two studies - data of a field experiment (Keppeler, 2023) and a follow-up pre-registered survey experiment (n>500 public recruiters, data collection ends in October 2023). The field experimental results show that both, human raters and AI applications, might discriminate against women, depending on the specific position. The survey experiment applies a 2x2 factorial design to understand the extent to which recruiters rely on algorithmic (compared to human) advice determining person-job fit, and in how far a debiasing intervention (compared to no intervention) can reduce potential biases against job candidates differing in both gender and ethnicity. The study contributes to the understanding of how AI applications shape decisions comparing public and private sector recruiters to foster the understanding of human-machine interaction and design policies preserving representative bureaucracies.
 
Authors (alphabetical):
Jana Borchert+, Florian Keppeler*, Vibeke Lehmann Nielsen*, Mogens Jin Pedersen#
 
Affiliations:
* Department of Political Science, Aarhus University (Denmark)
+ Department of Corporate Management and Economics, Zeppelin University (Germany)
# Department of Political Science, Copenhagen University (Denmark)
 

If you are not a CEPDISC fellow and wish to join pleace contact maj.carlsen@ps.au.dk