In a period of accelerated innovation—particularly in machine learning and artificial neural networks—the societal reach of AI-enabled technologies has greatly expanded, with far-reaching consequences for social organization, knowledge regimes, and global economic structures.

These consequences include: (1) the reshaping of power dynamics, as AI systems mediate decision-making processes previously governed by human judgment; (2) the intensification of surveillance and control mechanisms that contribute to new forms of social normalization and exclusion; (3) the destabilization of epistemic authority, as algorithmic models challenge traditional expertise and generate contested forms of knowledge production; and (4) the exacerbation of economic inequalities, as AI-enabled automation reshapes labor markets and asymmetrically redistributes wealth. (5) Taken together, these developments require a critical examination of the ethical, political, and social frameworks that determine the integration of AI technologies into democratic societies.

This seminar focuses on critically examining how AI systems not only reflect the world as it is, but also contribute to reproducing and exacerbating the very structures of inequality and exclusion they supposedly seek to overcome. Particular attention will be paid to algorithmic data processing in the context of platform capitalism, state surveillance, biometric control, and the increasing commodification of social practices.

To what extent do AI technologies encode categories of social difference—such as ethnicity, race, gender, class, disability, age, and sexual orientation—and how do these representational mechanisms affect social participation, justice, and epistemic authority?

A particular focus of the seminar is the critical analysis of central concepts that shape current understandings of AI and digital society: machine learning, dataveillance, datafication, digital ecology, algorithmic bias, and AI bias. These terms describe core processes through which data-driven technologies intervene in social, political, economic, and ecological systems—often under the guise of objective efficiency, but frequently with opaque and normatively charged effects.

Critical AI examines the institutional, economic, and political contexts in which AI applications emerge and the associated epistemic and social exclusions. Likewise, the perspective of queer AI opens up new conceptual spaces beyond binary classification logics. It critically engages with normative assumptions about gender, embodiment, identity, and representation embedded in data models and training datasets, while simultaneously proposing speculative alternatives to hegemonic AI designs.

Another focus is on data feminism and feminist AI, an approach that integrates feminist theory, postcolonial studies, social justice, and critical data practices. Data feminism challenges the illusion of objective data collection and advocates for contextual, participatory, and just forms of knowledge production. It foregrounds the visibility of structural inequalities and critically examines how epistemic violence is embedded in data models—through the exclusion, distortion, or homogenization of plural social realities. These approaches are complemented by postcolonial and decolonial perspectives on race and technology that examine how AI infrastructures perpetuate racialized classifications—for example, in biometric facial recognition, profiling, or algorithmic decision-making in policing and migration systems.

Our seminar is largely structured as a continuous assessment course. Through collaborative material analysis, we will examine and contextualize visual material together.

Each session will consist of a theory input (20 minutes), case studies (50 minutes), and close reading (20 minutes). Students are expected to attend regularly, participate in discussions, and prepare a short impulse presentation (5 minutes) during the semester.

Performance assessment will be completed with a written final paper (approximately 4-5 pages), with details discussed in the first session.