Tackling Algorithmic Disability Discrimination in the Hiring Process: An Ethical, Legal and Technical Analysis

Abstract

Tackling algorithmic discrimination against persons with disabilities (PWDs) demands a distinctive approach that is fundamentally different to that applied to other protected characteristics, due to particular ethical, legal, and technical challenges. We address these challenges specifically in the context of artificial intelligence (AI) systems used in hiring processes (or automated hiring systems, AHSs), in which automated assessment procedures are subject to unique ethical and legal considerations and have an undeniable adverse impact on PWDs. In this paper, we discuss concerns and opportunities raised by AI-driven hiring in relation to disability discrimination. Ultimately, we aim to encourage further research into this topic. Hence, we establish some starting points and design a roadmap for ethicists, lawmakers, advocates as well as AI practitioners alike.CCS Concepts: • Social and professional topics → People with disabilities; Governmental regulations; • Computing methodologies → Artificial intelligence.

Artificial Intelligence in the Auspices of Law: A Diverge Perspective

“…48 The formulation of these tenets necessitates interdisciplinary synergy, enlisting the expertise of legal luminaries, technocrats, ethicists, and policy architects, to guarantee that the regulatory framework remains synchronous with the unfolding contours of the AI landscape. 49 One of the cardinal considerations intrinsic to the establishment of a legal framework for AI pertains to the amelioration of inherent biases and discriminatory predilections within AI algorithms, AI systems are predicated upon erudition from historical data repositories, thereby susceptible to encoding prevailing biases inherent within societal frameworks. 50 The crafting of regulations mandating transparency, accountability, and perpetual auditability of AI algorithms emerges as an exigency of cardinal consequence to ameliorate these biases and ensure the even-handedness of decision-making U N I V E R S I T A S G A D J A H M A D A processes.…”

“…49 One of the cardinal considerations intrinsic to the establishment of a legal framework for AI pertains to the amelioration of inherent biases and discriminatory predilections within AI algorithms, AI systems are predicated upon erudition from historical data repositories, thereby susceptible to encoding prevailing biases inherent within societal frameworks. 50 The crafting of regulations mandating transparency, accountability, and perpetual auditability of AI algorithms emerges as an exigency of cardinal consequence to ameliorate these biases and ensure the even-handedness of decision-making U N I V E R S I T A S G A D J A H M A D A processes. Concomitantly, the regulatory schema must be adeptly calibrated to accommodate the distinct challenges occasioned by AI, embracing nuanced facets such as informed consent when interfacing with AI-derived insights and prognostications anchored in personal data.…”

Mapping the landscape of ethical considerations in explainable AI research

“…This effort aligns closely with the broader endeavor towards ethical governance of AI. Indeed, the ethical implications of AI technologies have gained significant attention due to their potential to perpetuate existing inequalities, produce unintended negative consequences, and create new ethical dilemmas Buyl et al, 2022;Jobin et al, 2019;Pastaltzidis et al, 2022). However, the extent to which XAI research genuinely addresses ethical considerations, and effectively assimilates them into the design, development, and evaluation of AI systems, remains a topic of considerable debate (Balasubramaniam et al, 2023;van Otterlo & Atzmueller, 2020;Kaur et al, 2020;Alufaisan et al, 2021;Chazette et al, 2019).…”

FairGridSearch: A Framework to Compare Fairness-Enhancing Models

“…Machine Learning (ML) models are increasingly utilized in critical decision-making applications, such as workforce recruiting [1], [2], justice risk assessments [3], [4], and credit risk prediction [5], [6]. Even though ML algorithms are not intentionally designed to incorporate bias, studies have shown that ML models not only reproduce existing biases in the training data [7] but also amplify them [8], [9], [10].…”