Opening the ‘black box’ of HRM algorithmic biases – How hiring practices induce discrimination on freelancing platforms
研究揭示了自由职业平台Upwork的排名算法如何通过工作完成数量间接导致对女性、黑人女性和亚裔求职者的歧视，为理解算法偏见提供了新视角。
• Uncovering an indirect discrimination mechanism explaining bias in algorithmic rankings. • Women freelancers are ranked lower by Upwork’s ranking algorithms than men. • Black women freelancers receive fewer jobs and are ranked lower than White women. • Black men freelancers are not disadvantaged compared to White men on freelancing platforms. • Inverted U-shaped mediation effect of age; mid-aged candidates are ranked highest. Online freelancing platforms extensively apply algorithms and AI, for example, to rank freelancers. These platforms are often considered neutral for not displaying freelancers’ gender, race, and age, but recent studies have revealed mounting freelancer complaints of unfair treatment and discrimination stemming from the platforms’ algorithms. Drawing from social dominance theory, this study contributes to the algorithmic HRM literature by uncovering an indirect algorithmic discrimination mechanism explaining bias in algorithmic rankings. By using an Upwork dataset of 44,167 freelancers and leveraging structural equation modeling, we find that the number of jobs completed through the platform mediates the effects of gender, race, and age on the platform’s ranking, demonstrating discrimination against female, Black women, Asian, and younger candidates. The study’s theoretical contributions to the algorithmic HRM literature, the methodological contribution of a novel AI picture analysis tool, and managerial implications for online freelancing platforms and HR departments are discussed.