Discriminated by an algorithm: a systematic review of discrimination and fairness by algorithmic decision-making in the context of HR recruitment and HR development

Abstract

Algorithmic decision-making is becoming increasingly common as a new source of advice in HR recruitment and HR development. While firms implement algorithmic decision-making to save costs as well as increase efficiency and objectivity, algorithmic decision-making might also lead to the unfair treatment of certain groups of people, implicit discrimination, and perceived unfairness. Current knowledge about the threats of unfairness and (implicit) discrimination by algorithmic decision-making is mostly unexplored in the human resource management context. Our goal is to clarify the current state of research related to HR recruitment and HR development, identify research gaps, and provide crucial future research directions. Based on a systematic review of 36 journal articles from 2014 to 2020, we present some applications of algorithmic decision-making and evaluate the possible pitfalls in these two essential HR functions. In doing this, we inform researchers and practitioners, offer important theoretical and practical implications, and suggest fruitful avenues for future research.

Ethics Guidelines for Using AI-based Algorithms in Recruiting: Learnings from a Systematic Literature Review

“…We summarized our categorization in Table 1. Köchling & Wehner [15]. stated, in HR recruitment and development there are two fairness types that need to be considered: Objective and subjective fairness perceptions of applicants and employees about the usage of algorithmic recruiting decisions.…”

“…These guidelines were published from technological companies e.g., Microsoft, Google or IBM as well as from governmental institutions e.g., EU Ethics guidelines for trustworthy AI. However, a comprehensive systematic literature review that provide structure for these guidelines for research and practice in the context of recruiting is missing as existing systematic approaches in this field are limited to certain journals [15].…”

In the Pursuit of Privacy: The Promises and Predicaments of Federated Learning in Healthcare

“…The WHO has recently released a guideline from various industry experts, academics, and public sector officials with an emphasis on the protection of human autonomy, equity, transparency, and sustainability which is indicative of a greater trend by the UN to encourage mindfulness in regard to ML (WHO, 2021). Potential biases that may arise due to disproportionate representation of minority groups is one of the endemic problems that arise from current ML, commonly referred to as "Algorithmic Discrimination" (Köchling and Wehner, 2020). For example, Google's facial recognition algorithm was widely criticized for its appalling identification of Black people as apes in 2015; they promptly "fixed" the issue by preventing the algorithm from classifying gorillas (Mulshine, 2015).…”

Word vector embeddings hold social ontological relations capable of reflecting meaningful fairness assessments

“…Thus, a vector space constructed using such documents will also represent such a bias (Caliskan et al 2017;Garg et al 2018). These biases have been seen as a hindrance to the effectiveness of using embeddings for social interaction applications, such as their use in candidate selection (Köchling and Wehner 2020). However, other biases inherent in Word Embeddings can in some instances be useful for extracting an underlying concept that has caused such a bias to manifest.…”