Conference Agenda

As public administrations increasingly adopt automated decision-making (ADM) to deliver services and allocate resources, a distinct form of structural discrimination has taken shape. Algorithmic discrimination, unlike traditional forms, is characterised by its scale, opacity, granularity, and constancy. These features present novel challenges for anti-discrimination law, which relies heavily on ex post remedies, individual complaints, and formal equality frameworks. The limitations of this model are becoming evident across a range of public sector domains—employment services, welfare distribution, healthcare triaging, and automated eligibility assessments—where algorithmic systems can exclude marginalised groups in invisible and systemic ways.

This paper argues for a regulatory and normative shift in how equality law addresses algorithmic discrimination, especially within public administration. It proposes an ex ante regulatory approach grounded in substantive equality, which enables institutions to prevent structural exclusion before harm occurs. Ex ante regulation refers to proactive obligations that alter the timing and nature of compliance—requiring actors who design, procure, or deploy ADM to identify and mitigate discriminatory impacts in advance. The goal is to move from reactive to anticipatory governance.

The paper situates this shift in the broader context of democratic governance and the evolving responsibilities of public institutions. Transparency and due process alone are insufficient to address the embedded and cumulative effects of algorithmic bias. Substantive equality provides a more context-sensitive framework to promote equal treatment, especially for historically disadvantaged groups. This reframing not only addresses harm but treats diversity as a resource, i.e., a foundation for designing fairer administrative systems that serve plural societies more effectively.

To build this framework, the paper draws on a comparative analysis of equality law across the European Union, India, and South Africa. The EU has been at the forefront of regulating emerging technologies, and its global leadership in AI governance—often described as the “Brussels Effect”—positions it as a key actor in shaping algorithmic accountability. However, its equality law remains shaped by a hybrid approach, in which positive action is treated as a narrow exception and formal neutrality remains the dominant logic. In contrast, India and South Africa treat affirmative action as a core component of equality, constitutionally recognising proactive duties to redress disadvantage. These jurisdictions offer valuable lessons for developing positive duties within equality law without abandoning commitments to fairness, transparency, or proportionality.

The paper introduces the concept of algorithmic affirmative action to describe the set of measures needed to implement substantive equality in ADM: use of protected characteristics in debiasing, inclusive training datasets, equality impact assessments, and institutional responsibility for fair outcomes. These measures are particularly relevant for public administration, where decision-making must not only be efficient but also inclusive, accountable, and democratically legitimate.

By embedding positive equality duties into public-sector algorithmic governance, the paper shows the way for law and policy to move beyond reactive control to proactive inclusion. This approach supports a model of administration that embraces diversity not as a challenge to be neutralised, but as an opportunity for democratic participation and equitable service delivery.

State-imposed literacy expectations on citizens in today’s increasingly digitalized public administrations extend far beyond the traditional understanding of literacy as the ability to read, do basic calculations, and write. The independent navigation of digital public services now requires citizens to demonstrate a broad range of competencies, such as digital literacy, information literacy, proficiency in the national language, the ability to identify errors in the system, or the capacity to seek help when encountering difficulties. Disparities in these skill sets result in access and engagement barriers of citizens with digital government. As interactions with public services become more standardized and less personalized, citizens who are unable to live up to those skill expectations, often those most reliant on public services, are at risk of marginalization and further exclusion.

Drawing on qualitative empirical data from interviews with Dutch professionals who support citizens struggling to navigate public services independently, this paper begins with examining which skill expectations the state places onto citizens. The data analysis follows a thematic coding approach (Clarke & Braun, 2006) and is contextualized within existing scholarship on literacy, citizenship and digital government. Building on this empirical foundation, the paper develops the theoretical concept of the ‘responsive citizen’, incorporating elements of grounded theory (Glaser & Strauss, 1967). The notion of the ‘responsive citizen’ in digital government reflects the shifting expectation that citizens must take responsibility for acquiring the necessary skills to engage effectively with public services instead of the government being responsible to make sure that all citizens are able to access services.

The idea of the ‘responsive citizen’ is developed in conversation with Martha Fineman’s legal feminist theory of the ‘responsive state.’ She advocates for a ‘responsive state’ that recognizes the universal and inherent vulnerability of all legal subjects, thereby establishing a “relationship of responsibility between the state and individual” (Fineman, 2010, p. 255). This responsibility entails that the state actively responds to the vulnerabilities of its citizens, for example, through robust support structures (2010) and would arguably also entail the responsibility of the state to ensure that all citizens can access public services, regardless of their skill abilities. Fineman’s (2010) advocacy for a ‘responsive state’ emerges from her critique of the neoliberal state, which prioritizes individual liberty above all else and constructs the legal subject as autonomous and self-sufficient, assumptions that are reflected in various aspects of today’s digital government.

In sum, this paper empirically investigates how the digitalization of the Dutch government has produced a range of skill and literacy expectations for citizens, fostering the emergence of a culture of the ‘responsive citizen’. Although it does not focus exclusively on the 'bright side' of digital public administration, this contribution aligns well with Study Group XXIII on Administration, Diversity, and Equal Treatment, as it highlights how a shift from the 'responsive citizen' to the 'responsive state' can enhance equitable access to and engagement with digital government services for citizens.

Administrative law plays an important role in fostering diversity as a driver for positive developments and to create a culture of gender equality and inclusion. However, one of the fields where its role has been little explored is its influence in the regulation and management of online gender-based hate speech to implement the visibility of diversity.

The study aims at analysing the recent phenomenon of online gender-based hate speech in Italy and France through a comparative legal and case law methodology, in order to see how the administrative bodies can implement some measures to let women participate in public life.

Online gender-based hate speech has, in fact, only recently become relevant for multiple reasons.

Firstly, there is no clear international legal framework on it. Different definitions of gender-based hate speech has been adopted at international, European and national levels, but they often overlap, or they present some “grey zones” where larger discretion is left to the States. Problematic profiles can also be identified in the so called “neutrality” of the online space; in the linguistic expression adopted; in the lack of a legal and linguistic shared definition of hate speech and sexist hate speech; finally, in the lack of legally binding instruments for the punishment of hate speech. Secondly, online gender-based hate speech can determine social effects in terms of participation of women in the democratic life. Research has shown that women are particularly vulnerable to hate speech online, with direct personal attacks. Certain factors must also be taken into account in the analysis of this phenomenon: the illusion of anonymity, the lack of perception of the target audience, the lack of awareness about the consequences of acting online, the persistence of content on platforms.

These considerations lead us to the third goal of this research: what is the role of administrative law, given that it has among its missions to ensure everyone their fundamental rights?

This contribution lastly aims to demonstrate how administrative law, through the collaboration between public and private sectors, plays a key role in the fight against gender-based hate speech online. Best practices of cooperation between the public and private sectors in Italy and France will be explored in order to see if they are able to overthrow and counteract online hate against women and to promote a so called “digital literacy”. In particular, the French Laboratory on women’s rights online will be analysed, especially for the profile regarding the cooperation between the Ministry of Foreign Affairs and civil society. On the other side, in Italy, the National Network for Combating Hate Speech will be examined, in order to explore its influence on the changes that have been made at administrative level.

The ultimate and more general aim of this research is to demonstrate that a powerful cooperation between different stakeholders is necessary not only to contribute to the creation of a stable democracy, but also to create a more pluralistic society.

The increased use of new technologies, or the innovative application of already existing technologies, is reshaping the dynamic between the welfare state and its citizens, disproportionately impacting marginalized individuals through punitive applications of technology, particularly artificial intelligence (AI). Sweden is often recognized as a near-perfect, inclusive, and egalitarian welfare system and existing scholarship has extensively analyzed the success of the social democratic welfare model. However, the Swedish welfare state is argued to have changed. It remains unclear if recent and future AI strategies for the public sector, which inevitably will shape the Swedish public administration, considers the needs of welfare recipients and aligns with the values attributed the Swedish welfare state.

Citizens are not a homogenous group; they have different skill sets and various abilities to understand and exercise their duties and rights. And thus, an automated administration should recognize this. Literature has nonetheless devoted limited attention to how citizens are included in the debate on how to reshape the digital transformation in the public sector. This paper addresses this gap by investigating if and how citizens are being represented in policy documents related to AI, as well as how policy makers and actors in charge consider the needs and perspectives of citizens who will be the most affected by it. An issue we conceptualize as governance blindness. Governance blindness does not only constitute a fairness issue, but from a broader perspective, it also represents a democratic issue. This as certain segments of the population are then consistently and continuously further marginalized.

This paper seeks to contribute to the knowledge on the role of citizens in local government AI use by investigating if and how citizens are being represented, in policy documents, as well as how policy makers consider the needs and perspectives of citizens. More specifically, this paper explores the AI strategy of Region Skåne and Region Örebro. In order to develop the concept of governance blindness in public administration and more specifically, in the specific case of AI strategies, this paper adopts the following methods. This paper is interested in the following research question: (How) are citizens considered and represented in Swedish regional AI strategies and what vision of AI underlies these documents? In order to answer this question, we conduct a policy analysis drawing on Asdal and Reinertsen’s (2022) practice-oriented approach to document analysis. Our selected material consists of regional AI strategies as well as other policy documents related to these strategies. By doing so, this study develops a deeper understanding of the evolving state-citizen dynamic, with a particular focus on marginalized and vulnerable populations.