Queering Law and Artificial Intelligence: Towards a Technology Oriented Principle of Equality

Artificial Intelligence (AI) is increasingly failing minorities across multiple domains, including government welfare decision-making, airport security control, policing and criminal justice, online content moderation, hiring processes, and healthcare decisions. These failures disproportionately impact individuals based on ethnicity, sex, gender, sexual orientation, age, socioeconomic status and other intersecting identities. While AI is frequently described as a «mirror» that reflects human behavior, this analogy oversimplifies its role. In reality, AI risks exacerbating societal inequalities because of its capacity to establish and entrench new patterns of discrimination, and due to a social and legal framework that is not often well equipped to respond to the “mathematization of inequality”. Rather than being a simple mirror, AI acts as a magnifying lens: on one hand, it reveals with clarity the distortions embedded in human behavior and institutional practices; on the other, it risks amplifying and perpetuating these distortions to an alarming degree. Law faces an unprecedented challenge: ensuring the preservation of fundamental rights in a context where artificial intelligence not only interacts with individuals but also shapes behaviors, decisions, and societal values. In this scenario, law can no longer consider AI merely as an object of regulation or as a neutral tool. Instead, AI must be recognized as a subject that, while shaped by humans, also exerts a transformative influence on human behavior and the underlying social paradigms. This shift in perspective mandates a reflection on the interaction between AI and constitutional law and principles. Equality and antidiscrimination law, traditionally understood, may prove insufficient in addressing the new forms of discrimination generated by automation and algorithmic decision-making. It becomes essential to explore how the concept of equality can be adapted to the technological challenges, taking into account not only individual rights but also the systemic relationships and structural mechanisms that affect the distribution of opportunities and resources. As a matter of fact, much has been said in academic discourse about the discriminatory potential of AI and about the failures and shortcomings of the traditional conception of equality and antidiscrimination law in remedying these discriminations. Now, it is time to respond to these challenges by proposing potential normative solutions. This thesis seeks to fill this gap. In particular, the incongruity of three dimensions has become evident. First, equality as an anticlassification rule; second, the static and binary conception of social categories such as sex, gender, and sexual orientation; and third, the technological binarism that underpins algorithmic reasoning. These dimensions reveal a fundamental tension: while anticlassification approach to equality aims to neutralize distinctions between individuals, the rigidity of binary social categories and the deterministic logic of algorithms reinforce those very distinctions. The study presented here aims to construct a technology oriented principle of equality capable of addressing this triple challenge. To achieve this, the arguments underpinning the construction of this principle are grounded in three axioms. First, there is a need to move beyond mere formalism and embrace a concept of substantive equality that, alongside the allocative dimension of nondiscrimination, also recognizes the importance of recognition of different identities, participation and representation as means of empowerment for minorities, and the transformation of society through the accommodation of differences. Second, this thesis argues for the necessity of acknowledging the social construction of categories such as sex, gender, and sexual orientation also through legal rules. Recognizing this dimension implies assigning the law a potential role in shaping society and technology to foster substantive equality. By understanding these categories as socially constructed, law and technology can serve both as a transformative tool to challenge structural inequalities and promote inclusivity. Finally, AI must be recognized both as a tool and a target of regulation. In this sense, the technology oriented principle of equality shall be conceived as an ex ante instrument that complements the classic remedial function of antidiscrimination law traditionally exercised by courts. It is important to emphasize from the outset that, while this investigation is confined to exploring the intersections of discrimination, AI, and sexual and gender minorities, the technology oriented principle of equality aims to address challenges faced by all minorities today. These challenges arise from various and intersecting identity traits, including ethnicity, socio-economic conditions, disability, age, and more. As a matter of fact, a queer discourse is not limited by thematic boundaries. Instead, it operates as a methodological tool capable of producing a transformative effect on all categories fixed by law. By questioning the rigidity of these categories, a queer perspective fosters a dynamic and inclusive approach that seeks to dismantle structural inequalities across all dimensions of identity, extending the relevance and applicability of the principle to broader contexts. Starting from these assumptions, the quest to find a technology oriented principle of equality unfolds across two parts. Part I, titled Equality and Antidiscrimination Law, comprises Chapters 1, 2, and 3. This section lays the theoretical foundations necessary to understand the challenges of addressing discrimination. Chapter 1 explores the multidimensional reconstruction of equality, emphasizing the limitations of formal equality and advocating for a substantive conception that addresses historical and systemic disadvantages faced by certain social groups. It examines how constitutional principles of equality justify and legitimize antidiscrimination legislation and should guide both legislative power and the judiciary in promoting substantive equality. Chapter 2 analyzes the main features of antidiscrimination law from a comparative perspective, focusing on direct discrimination, indirect discrimination, and affirmative action. It reveals how different legal models approach these concepts, highlighting the need to move beyond intentionality and binary comparators to recognize systemic and unconscious forms of discrimination. This chapter underscores the importance of addressing not just individual acts of discrimination but also broader structural inequalities. Chapter 3 introduces queer legal theories as a critical lens to examine the limitations of traditional legal frameworks in addressing discrimination based on sex, gender, and sexuality. It critiques the static and binary conception of these categories in law, emphasizing the social construction of identities. By adopting a queer theoretical perspective, this chapter sets the stage for understanding how existing antidiscrimination laws may fall short in protecting gender minorities and queer people. Part Two, titled Towards a Technology oriented Principle of Equality, encompasses Chapters 4 and 5. In this section, the theoretical foundations established in Part One are utilized to evaluate the limitations of antidiscrimination law when confronted with AI-based queer discrimination and to construct a technology oriented principle of equality grounded in the theoretical framework previously discussed. Chapter 4 critically examines the intersection of artificial intelligence, queer theories, and antidiscrimination law. By framing technical and sociotechnical biases, it highlights how AI systems can perpetuate and even exacerbate forms of allocative and symbolic discriminations, particularly against sexual and gender minorities. The chapter argues that traditional antidiscrimination laws are often insufficient to address the unique challenges posed by AI due to their reliance on formal equality and ex post facto remedies. The analysis will be conducted by looking at three case-studies of AI-based discrimination towards queer minorities, dealing with automated recognition technology, AI in the healthcare sector, and algorithmic online contend moderation. Eventually, the chapter advocates for a shift towards an ex ante, substantive approach that integrates equality principles directly into the design, development and deployment of AI systems. Chapter 5 proposes a multi-layered framework for a technology oriented principle of equality. It outlines three legal principles, “equality by design”, “equality by participation” and “equality by education” to address both technical and sociotechnical AI biases. By analyzing different regulatory models, the chapter builds a principle of equality through a sociotechnical approach that combines technical safeguards with participatory mechanisms and educational initiatives. This integrated framework aims to create an AI landscape that not only avoids perpetuating discrimination but actively promotes equality by embedding safeguards into the very fabric of AI technologies and its social and regulatory environment so that AI can become tool of equality rather than instruments of power asymmetry, subordination and discrimination.