A Framework for Fairness: A Systematic Review of Existing Fair AI Solutions

Abstract

In a world of daily emerging scientific inquisition and discovery, the prolific launch of machine learning across industries comes to little surprise for those familiar with the potential of ML. Neither so should the congruent expansion of ethics-focused research that emerged as a response to issues of bias and unfairness that stemmed from those very same applications. Fairness research, which focuses on techniques to combat algorithmic bias, is now more supported than ever before. A large portion of fairness research has gone to producing tools that machine learning practitioners can use to audit for bias while designing their algorithms. Nonetheless, there is a lack of application of these fairness solutions in practice. This systematic review provides an in-depth summary of the algorithmic bias issues that have been defined and the fairness solution space that has been proposed. Moreover, this review provides an in-depth breakdown of the caveats to the solution space that have arisen since their release and a taxonomy of needs that have been proposed by machine learning practitioners, fairness researchers, and institutional stakeholders. These needs have been organized and addressed to the parties most influential to their implementation, which includes fairness researchers, organizations that produce ML algorithms, and the machine learning practitioners themselves. These findings can be used in the future to bridge the gap between practitioners and fairness experts and inform the creation of usable fair ML toolkits.

A seven-layer model with checklists for standardising fairness assessment throughout the AI lifecycle

“…While AI fairness checklists are important in organisations to formalise ad-hoc processes, the checklists need to be aligned with the existing workflows to avoid their misuse [62]. Richardson and Gilbert [63] reviews various popular software toolkits and checklists as the solutions to tackle algorithm bias. Seedat et al [64] provides checklists at the four stages of the ML pipeline: Data, Training, Testing, and Deployment.…”

“…Thus, systematic selection of appropriate variables and collection of relevant data is necessary to reduce algorithm bias [63].…”

Identifying Sociotechnical Harms of Algorithmic Systems: Scoping a Taxonomy for Harm Reduction

“…Allocative harms encompass problems arising from how algorithmic decisions are distributed unevenly to different groups of people [22,162]. These harms occur when a system withholds information, opportunities, or resources [22] from historically marginalized groups in domains that affect material well-being [127], such as housing [41], employment [176], social services [15,176], finance [100], education [102], and healthcare [139].…”

Fairness in Agreement With European Values: An Interdisciplinary Perspective on AI Regulation

“…Other research examples, once informed on practitioners' needs, focused on designing different AI fairness solutions: checklists to be aligned with teams' workflows and organizational ad-hoc processes, fairness frameworks or internal algorithmic auditing protocols designed for industrial applications [61,91]. Recently, Richardson and Gilbert [97] proposed a complete industry framework of stakeholders and fairness recommendations while specifying operationalization pitfalls. Ibáñez and Olmeda [47] distinguished two main perspectives on operationalizing fairness practices in organizations: a bottom-up, reactive approach, where prior organizational processes restrain best practices, or top-down, where a proactive approach is set in place according to the translation of principles and methods as actionable, iterative steps designed with stakeholders' needs and concerns in mind.…”