source 7.3

Original Research Overview

• Title: Discriminated by an Algorithm: A Systematic Review of Discrimination and Fairness by Algorithmic Decision-Making in the Context of HR Recruitment and HR Development

• Authors: Alina Köchling, Marius Claus Wehner

• Published: November 20, 2020

Abstract

• Context: Algorithmic decision-making is increasingly utilized in human resource (HR) recruitment and development.

• Purpose: To analyze potential discrimination and unfair treatment caused by algorithms in HR processes.

• Methodology: Systematic review of 36 journal articles from 2014 to 2020.

• Findings: Discusses applications, pitfalls, implications, and suggests future research directions in HRM context concerning fairness and discrimination.

Introduction

• Emergence of Algorithmic Decision-Making: Increasingly used for its efficiency and objectivity in HR functions due to digitalization.

• Challenges: Potential for implicit discrimination and perceived unfairness against certain groups remains poorly researched in HRM.

• Goals:

  • Clarify current research state.

  • Identify gaps.

  • Provide future research directions.

Definition and Importance of Algorithmic Decision-Making

• Definition: Automated decision-making and standardization of routine workplace decisions (Möhlmann and Zalmanson, 2017).

• Potential Benefits:

  • Discover hidden talents.

  • Review applications at scale (Silverman and Waller, 2015).

• Market Trends: 79% of AI specialists view AI as essential for competitive advantage (Deloitte, 2020).

• Adoption: Companies like Google, IBM, and Vodafone utilize algorithmic systems for HR tasks (Daugherty and Wilson, 2018).

Economic Motivations for Algorithmic Use

• Cost-saving: Firms utilize algorithms to reduce time and cost within HR processes.

• Bias Reduction: Intended to mitigate human biases, increase consistency, and fair treatment in hiring and development (Langer et al., 2019).

• Risks: Algorithms can perpetuate biases if trained on flawed data (Kim, 2016; Barocas and Selbst, 2016).

Bias and Discrimination in Algorithmic Decision Making

• Types of Risks:

  • Discrimination against groups based on gender, age, ethnicity (Arrow, 1973).

  • Algorithms may produce biased outcomes if trained on inaccurate or unrepresentative data.

• Case Study:

  • Amazon's hiring algorithm favored male candidates, leading to its discontinuation (Dastin, 2018).

Perceptions of Fairness and Transparency

• Applicant Perspectives: Acceptance of algorithmic systems is often lower compared to traditional human practices (Kaibel et al., 2019).

• Discrepancy in Views: Enthusiasm over algorithmic efficiency contrasted by concerns over discrimination and unfairness.

• Need for Future Research: Insights on fairness perceptions are vital for developing responsible HR practices.

Methodology of Systematic Review

• Process: Systematic reviews adhere to PRISMA standards, covering search terms, screening, and inclusion criteria.

• Database Search: Employed social science citation index (SSCI) and EBSCO Business Source for a comprehensive review.

• Findings: Identification of key patterns in the literature concerning algorithmic bias and discrimination.

Key Findings

• Main Focus of Literature: Majority target bias, discrimination, and fairness.

• Interdisciplinary Nature: Research spans across management, computer science, law, and psychology.

• Need for Empirical Evidence: Call for more quantitative studies and focus on perceived fairness in algorithmic HR systems.

Implications for HRM Practice

• Navigating Algorithmic Implementation:

  • Awareness of potential biases and accountability for algorithmic decisions in hiring.

  • Need for transparency in data usage and algorithmic processes.

  • Employing human oversight alongside automated systems can enhance fairness.

Conclusion

• Research Conclusions: Understanding and addressing biases in algorithmic decision-making is critical as organizations increasingly rely on these systems.

• Future Research Directions:

  • Explore perceptions of fairness among employees.

  • Investigate the compatibility of algorithmic and human decision-making processes in HR recruitment and development.