Hugo Cossette-Lefebvre

Structural discrimination is sometimes presented as a unique subtype of discrimination. Yet, it is unclear how this concept should be defined. I argue that structural discrimination against a group G obtains when (1) G is a social group in society S; (2) members of G typically experience enduring, cross-cutting comparative disadvantages that are plural, multi-facetted, and that affect different significant spheres of life; and (3) the comparative disadvantages are created by the combined effects of different interdependent social practices. I show how structural discrimination is distinct from individual instances of direct and indirect discrimination and that this concept is different from the notion of structural injustice.

In this paper, I argue that AI systems can wrong persons based on disposition-like states or quasi-dispositions they hold. Persons can be wronged by AI systems not only by their observable behaviour, but also in how these systems reach certain con-clusions or due to the underlying quasi-dispositions driving their behaviour. I build on the literature on doxastic wrongs and the norms regulating belief formation to argue that even if AI systems do not hold beliefs, as they are not conscious entities, the disposition-like states of AI systems should be subjected to moral evaluation given that they are integrated into socio-political relations. In closing, I explore de-sign and deployment implications. This analysis highlights the necessity of moving beyond harm-based assessments to develop ethical AI systems.

This chapter explores the impact of recommender systems on autonomy. We draw on feminist relational approaches to argue that recommender systems can threaten autonomy in four key ways. (i) Recommender systems limit people’s exposure to options, creating tunnel vision. (ii) They undermine the imaginative capacities crucial to the critical reflection required for autonomy and constrain the options people take to be feasible for them. (iii) They can generate a “lens” through which people experience the world, reinforcing oppressive stereotypes and negatively affecting self-trust or self-respect. (iv) They create echo chambers that generate an epistemological problem for autonomy by privileging false beliefs. After identifying these threats, we argue that recommender systems are not inherently autonomy-undermining and could be designed to promote autonomy. To achieve this, we propose three design goals: (i) ensuring exposure to diverse content to prevent tunnel vision; (ii) safeguarding self-respect and avoiding the exploitation of individual vulnerabilities; and (iii) generally promoting truthfulness.

This paper investigates the relationship between discrimination, epistemic injustice, and social structures as it pertains to the marginalization of disabled people. Doing so, we argue for a new understanding of structural discrimination. While discrimination is often viewed as an agent-centered or rule-based phenomenon, we argue that a broader socio-structural lens is crucial for understanding how it emerges and persists over time. Drawing on the Human Variation Model (HVM) of disability, we analyze three subtypes of epistemic injustice—testimonial injustice, hermeneutical injustice, and non-disabled ignorance—and demonstrate their critical role in perpetuating structural discrimination. Our analysis addresses three gaps in the literature: (1) it clarifies the relationship between epistemic injustice and (structural) discrimination; (2) it highlights how epistemic injustices contribute to discrimination against disabled people; and (3) it allows us to have a better grasp of what structural discrimination is and how it differs from its direct and indirect counterparts.

Discrimination is typically understood to be either direct or indirect. However, we argue that some cases that clearly are instances of discrimination are neither direct nor indirect. This is not just a logical taxonomical point. Highly salient, contemporary cases of algorithmic discrimination – a form of discrimination which was not around (or, at least, not conspicuously so) when the distinction between direct and indirect discrimination was originally articulated – are best construed as a third form of discrimination – non-direct discrimination, we shall call it. If we are right, the dominant dichotomous distinction between direct and indirect discrimination should be replaced by our tripartite distinction between direct, indirect, and non-direct discrimination. We show how non-direct discrimination covers not only important types of algorithmic discrimination, but also allows us to make sense of some instances of implicit bias discrimination.

Many authors adopt a relational egalitarian outlook to explain why discrimination is wrongful. Roughly, for relational egalitarians, (social) equality is about whether all are treated and regarded as equals in society. However, this raises an important puzzle: how can relational egalitarians explain the wrong of indirect discrimination, which captures instances of differential impact absent differential treatment? In this paper, I consider three arguments relational egalitarians have put forward to respond to this question: (1) the compounding argument; (2) the negligence argument; and (3) the no-such-thing-as-indirect-discrimination argument. I argue that none is fully convincing. In closing, I offer a positive proposal: indirect discrimination is wrong either when (1) it contributes to the structural disadvantaging of a group in society by tracking a basing trait, or (2) when a particular practice amounts to wrongful negligence. I argue that each is sufficient to say that a practice is wrongfully indirectly discriminatory.

Is it wrong to discriminate against a fictive person? We show that prominent accounts of the wrongness of discrimination provide conflicting answers to this question. This raises the question: must an account of the wrongness of discrimination, to be extensionally adequate, imply that discriminating against a fictive person is wrongful? We argue that answering “no” to this question either proves difficult or comes with significant costs. In this way, our investigation helps to address the broader, but neglected question: can wrongful (direct) discrimination occur even when no actual person is directly targeted? Our investigation thereby ultimately helps to determine the plausible scope of anti-discrimination theories.

This chapter considers the emerging third wave of AI ethics. The third wave relies on the idea that AI should be approached structurally to fully appreciate its ethical implications. Two main tendencies emerge in contemporary discourses on just AI in this third wave. First, some insist on the importance of deliberative inclusion for the development, regulation, and deployment of AI. Second, many build on contemporary approaches in social philosophy to ensure that AI is not only consistent with but fosters and promotes social equality and personal autonomy in society. However, I argue that these two tendencies are not necessarily aligned with one another.

In this introduction, I briefly summarize Sophia Moreau's Faces of Inequality. I situate her monograph within two highly contemporary bodies of literature — relational egalitarianism and discrimination theory — to show how it provides important insights for understanding both what it means to treat others as equals in society and how to define wrongful discrimination. Moreau's work on discrimination is of great relevance for philosophers and socio-legal theorists alike as the commentaries from the symposium contributors demonstrate, including Dale Smith, Pablo Gilabert, Andrea Sangiovanni, Daniel Viehoff, Jessica Eisen, Alysia Blackham, and Iyiola Solanke.

Herjeet Kaur Marway recently proposed the Principle of Procreative Justice, which says that reproducers have a strong moral obligation to avoid completing race and colour injustices through their selection choices. In this article, we analyze this principle and argue, appealing to a series of counterexamples, that some of the implications of Marway's Principle of Procreative Justice are difficult to accept. This casts doubt on whether the principle should be adopted. Also, we show that there are some more principled worries regarding Marway's idea of a strong pro tanto duty not to complete injustices through one's procreative choices. Nonetheless, we believe Marway's arguments point in the right general direction regarding duties and structural injustice. Thus, in the final part, we suggest a positive proposal on how it would be possible to respond to the cases we raise. More specifically, we explore the suggestion that agents have a pro tanto duty to participate in eliminating structural injustice. Importantly, this duty can be satisfied, not only in procreation choices but in multiple ways.

Heloise Robinson argues that pregnant women have a higher moral status than non-pregnant persons and that, for this reason, pregnant women ought to be treated ‘noticeably’ better than non-pregnant persons.1 In this commentary, we present two challenges to Robinson’s argument. First, the compounding disadvantage objection: treating involuntarily, non-pregnant women worse than voluntarily pregnant women unjustly compounds their disadvantage. Second, the identity objection: treating non-pregnant people worse than pregnant people amounts to pro tanto wrongful discrimination based on a fundamental aspect of people’s identity. We conclude that either pregnant persons do not have a higher moral status than non-pregnant persons, or they do, but higher moral status so grounded does not justify better treatment. According to the standard view of moral status: (1) most human beings are persons; (2) persons have a higher moral status than non-persons; and (3) persons have an equal moral status. Many believe: (1) is true because most human beings have a sufficiently high level of personhood-generating psychological capacities; (2) is true because moral status is tied to personhood; and (3) is true because variations in the level of personhood-generating psychological capacities do not matter to an individual’s moral status once that individual’s level surpasses the threshold required for personhood.2 (3) is significant …

The status of indirect discrimination is ambiguous in the current literature. This paper addresses two contemporary and related debates. First, for some, indirect discrimination is not truly a distinct kind of discrimination, but it is simply a legal construct designed to address distributive inequalities between groups. Second, even if one accepts that indirect discrimination is a distinct type of discrimination, the connection between the two kinds of discrimination, direct and indirect, is debated. For some, they are distinct act-types, while for others, indirect discrimination should be conceived as a side effect of prior cases of direct discrimination. In this paper, I argue that indirect discrimination is a distinct act-type that can take place without being connected to prior instances of direct discrimination.

Whether it is about Québec independence, French language or immigration, nationalism is a crucial feature of Québec politics. The Québec 2018 election is not an exception. Scholars have developed theories about individual identity, the nation and nationhood, but we lack a citizens’ perspective. We provide the first thorough description of Quebeckers’ nationalism, which reveals a roughly normal (i.e. non-polarized) distribution of ethnic nationalism attitudes. Most importantly, we measure ethnic nationalism with a never tested measure in Québec and we show that it substantially explains vote choice in the Québec 2018 election—especially the support for the Coalition Avenir Québec. Our research builds bridges between debates in political philosophy and political science and deepens our understanding of the Québec 2018 election.

The use of predictive machine learning algorithms is increasingly common to guide or even take decisions in both public and private settings. Their use is touted by some as a potentially useful method to avoid discriminatory decisions since they are, allegedly, neutral, objective, and can be evaluated in ways no human decisions can. By (fully or partly) outsourcing a decision process to an algorithm, it should allow human organizations to clearly define the parameters of the decision and to, in principle, remove human biases. Yet, in practice, the use of algorithms can still be the source of wrongful discriminatory decisions based on at least three of their features: the data-mining process and the categorizations they rely on can reconduct human biases, their automaticity and predictive design can lead them to rely on wrongful generalizations, and their opaque nature is at odds with democratic requirements. We highlight that the two latter aspects of algorithms and their significance for discrimination are too often overlooked in contemporary literature. Though these problems are not all insurmountable, we argue that it is necessary to clearly define the conditions under which a machine learning decision tool can be used. We identify and propose three main guidelines to properly constrain the deployment of machine learning algorithms in society: algorithms should be vetted to ensure that they do not unduly affect historically marginalized groups; they should not systematically override or replace human decision-making processes; and the decision reached using an algorithm should always be explainable and justifiable.

Les démocraties s’articulent autour de deux pôles qui ne se réduisent pas l’un à l’autre, mais qui doivent communiquer constamment : la société civile et l’État. Or, l’équilibre à rechercher entre ces deux éléments est mis à mal par la consolidation de l’État administratif qui tend à autonomiser son action par rapport au reste de la société ; la société civile y est d’ordinaire exclue des prises de décisions. Cette exclusion peut toutefois être considérée positivement ou négativement. Comme il sera discuté à partir des philosophies de John Dryzek et d’Iris Marion Young, on devrait tenter de penser à la fois l’inclusion de la société civile dans certaines prises de décisions étatiques et son indépendance pour favoriser sa dimension critique envers l’État. De là, l’État administratif peut être considéré simultanément comme une opportunité de démocratisation et comme une évolution de l’État à surveiller et à critiquer.