The third meeting was led by the University of Manchester (UoM) and focused on feminist pedagogy and AI, with particular attention to how feminist theory can inform both critiques of AI systems and approaches to AI education. The session was organized into three interconnected parts: (1) historicising feminist critiques of AI, (2) outlining key principles of feminist pedagogy, and (3) discussing implications for educational practice.

The session brought together different but complementary perspectives on AI, feminism, and education. It began by clarifying that the purpose was not to define a fixed feminist AI curriculum, but rather to explore feminist traditions and critiques as resources for thinking about what a feminist AI pedagogy could look like. Two broad strands running through feminist responses to AI were highlighted: one centered on agency and design, and another focused on ethics and justice. Throughout the discussion, feminist critiques were presented not only as questions of representation, but also as examinations of the historical, political, and epistemological foundations of AI itself.

One part of the discussion focused on critiques related to representation and participation in AI. Feminist scholars have long questioned why AI development remains dominated by white, male perspectives and why marginalized groups are often excluded from positions of power within the field. Closely connected to this was the issue of biased datasets and algorithms. The session explored how AI systems are trained on historical data that already reflect existing inequalities and dominant worldviews, often reproducing gendered, racialized, and colonial biases. Several influential feminist and critical works on AI, data, and algorithmic oppression were referenced throughout the discussion.

The conversation then moved toward more foundational critiques of AI. Some feminist perspectives were presented as viewing AI not simply as a neutral technology shaped by biased data, but as a project historically connected to militarization, colonialism, managerialism, optimization, and automation. AI was discussed as emerging from longer histories of psychometrics, eugenics, and systems of social control. From this perspective, the problem is not only biased outputs but also the deeper assumptions and structures underlying AI systems. This led to important tensions within feminist debates: while some approaches focus on improving representation and rebalancing data, others question whether AI can truly be separated from the historical systems of power within which it developed.

At the same time, the session also acknowledged more optimistic feminist and posthuman approaches to AI. AI and digital technologies were discussed as potentially disrupting traditional boundaries between human and machine, nature and technology, and creating possibilities for new forms of identity, agency, and relationality. Feminist engagements with AI were therefore presented as diverse and sometimes contradictory, ranging from strong critiques of AI as a technology of domination to attempts to reclaim and redesign AI systems through feminist principles. Questions of labor and exploitation also played an important role throughout the discussion. AI systems were described not only as technical systems, but also as part of broader global infrastructures of extraction, discipline, and inequality. Particular attention was given to how AI development relies on forms of labor that are often invisible, precarious, and disproportionately carried out by racialized and gendered workers.

The second part of the session shifted toward feminist pedagogy. Feminist pedagogy was introduced as an approach fundamentally concerned with the relationship between teaching, learning, social justice, and liberatory social change. Drawing on feminist, anti-racist, and critical pedagogical traditions, the discussion emphasized that education is never neutral, but always shaped by broader social and political structures. Several key principles of feminist pedagogy were discussed. These included examining the relationship between power and knowledge, understanding knowledge as situated and partial rather than universal and neutral, foregrounding embodiment and emotion in learning, fostering reflexivity and collective critical reflection, and approaching learning as fundamentally relational. Teaching and learning were described not simply as processes of transferring information, but as practices deeply connected to questions of power, exclusion, and participation.

The discussion also emphasized the importance of situated knowledges. Knowledge production was described not as an objective “view from nowhere”, but as something always produced from specific social and embodied positions. Feminist pedagogy, therefore, encourages both teachers and students to reflect on whose perspectives are centered, whose experiences are excluded, and how learners themselves participate in the construction of knowledge.

These pedagogical ideas were then connected directly to technical education and computer science. The discussion highlighted critiques suggesting that computer science often operates through forms of abstraction that separate technical systems from their broader social and political contexts. Feminist approaches challenge this separation by encouraging students to think critically about how technical systems are embedded within social relations, infrastructures, and structures of power. Rather than treating technical competence as isolated from ethical or political questions, the session emphasized the importance of integrating critical and social perspectives into technical education itself.

An important distinction was also raised between two different approaches to feminist AI. The first asks how AI systems could be designed from the ground up based on feminist principles. The second assumes that existing systems already exist and instead focuses on mitigating harms such as bias after the fact. The discussion emphasized that the second approach risks becoming reductive because it addresses individual technical problems without challenging the deeper assumptions and structures shaping AI systems from the beginning. In response, the session reflected on the importance of understanding AI as a layered “stack”, where interventions may occur at different levels, from infrastructure and foundation models to community-based applications and datasets. This led to a broader discussion about both the possibilities and the limitations of feminist interventions within existing AI systems.

The session concluded by raising a series of broader questions for future work. These included questions about what a feminist AI curriculum should look like, who it is for, whose agency it amplifies, how curriculum design itself can become a feminist ethical practice, and how feminist pedagogy can avoid being captured or instrumentalized within data-driven educational systems. Rather than offering definitive answers, the session framed these as ongoing collective questions that require interdisciplinary collaboration, critical reflection, and continued dialogue.