On Consequentialism and Fairness

Abstract

Recent work on fairness in machine learning has primarily emphasized how to define, quantify, and encourage “fair” outcomes. Less attention has been paid, however, to the ethical foundations which underlie such efforts. Among the ethical perspectives that should be taken into consideration is consequentialism , the position that, roughly speaking, outcomes are all that matter. Although consequentialism is not free from difficulties, and although it does not necessarily provide a tractable way of choosing actions (because of the combined problems of uncertainty, subjectivity, and aggregation), it nevertheless provides a powerful foundation from which to critique the existing literature on machine learning fairness. Moreover, it brings to the fore some of the tradeoffs involved, including the problem of who counts, the pros and cons of using a policy, and the relative value of the distant future. In this paper we provide a consequentialist critique of common definitions of fairness within machine learning, as well as a machine learning perspective on consequentialism. We conclude with a broader discussion of the issues of learning and randomization, which have important implications for the ethics of automated decision making systems.

Delphi: Towards Machine Ethics and Norms

“…Other studies have focused on NLP applications with ethical motivations, such as cataloguing and detecting implicit social biases Zhao et al, 2021b;Blodgett et al, 2020). These works are broadly situated in the dominion of computational ethics (Card & Smith, 2020), and are predated by earlier logic programming approaches (Berreby et al, 2015;Pereira & Saptawijaya, 2007). We note a separate but critical line of work which inquires about the ethics of developing NLP technology itself (Leins et al, 2020;Tsarapatsanis & Aletras, 2021;Chubb et al, 2021).…”

Five sources of bias in natural language processing

“…Moving away from pure performance metrics and looking at the robustness and behaviour of the model in suites of specially designed cases can add further insights (Ribeiro et al, 2020). Card and Smith (2020) explore constraints to be specified on outcomes of models. Specifically, these constraints ensure that the proportion of predicted labels should be the same or approximately the same for each user group.…”

Case Study: Deontological Ethics in NLP

“…To more rigorously determine what is right and wrong, we rely on ethical theories. Card and Smith (2020) present an analysis of ethics in machine learning under a consequentialist framework. This paper is a kindred spirit in that we both seek to make a philosophical theory of ethics concrete within machine learning and NLP, yet the methods of the paper are somewhat orthogonal.…”

“…This paper is a kindred spirit in that we both seek to make a philosophical theory of ethics concrete within machine learning and NLP, yet the methods of the paper are somewhat orthogonal. Card and Smith (2020) provide a comprehensive overview of how the particular nature of consequentialist ethics is relevant to machine learning whereas we intend to provide tangible examples of how deontological ethical principles can identify ethically important areas of research. Saltz et al (2019); Bender et al (2020) advocate for explicitly teaching ethical theory as a part of machine learning and NLP courses; the case studies in this paper would be a logical extension of the material presented in such a course.…”