If there is one concept that connects all four critical traditions explored in this series, it is intersectionality, the recognition that social positions, forms of oppression, and systems of power do not operate independently but interact, reinforce, and modify each other in complex ways.

Kimberlé Crenshaw coined the term in 1989 to describe a specific legal problem: anti-discrimination law in the United States treated race discrimination and sex discrimination as separate categories, which meant that Black women, who experienced both simultaneously, in ways that could not be reduced to either alone, often fell through the gaps of legal protection. The concept was theoretical but its origin was practical: people at the intersection of multiple subordinated identities were being failed by frameworks built around single-axis analysis.

Intersectionality has since been extended far beyond its origins in legal theory. It now provides a framework for understanding how multiple systems of power, racial capitalism, patriarchy, heteronormativity, ableism, colonialism, interact to produce the specific experiences of people at their various intersections.

Applied to AI, intersectionality insists on a fundamental methodological commitment: disaggregated analysis. It is not sufficient to ask whether an AI system produces equitable outcomes across racial groups, or across gender groups, separately. We must ask how it performs at the intersection of race and gender, race and class, gender and disability, and other relevant dimensions. Systems that appear equitable on each single dimension can be profoundly inequitable at their intersections.

The four frameworks through an intersectional lens:

Critical Race Theory developed the concept of intersectionality and remains its primary institutional home. But CRT’s intersectional analysis is most fully developed at the intersection of race and gender, other axes of identity receive less systematic attention within CRT as traditionally conceived.

Feminist theory, particularly socialist and intersectional feminism, extends the analysis to the intersections of gender with class, race, sexuality, disability, and nationality. It draws attention to how feminist analysis that focuses on white middle-class women’s experiences has historically failed women at multiple intersections.

Marxist theory provides the economic dimension that intersectional analysis requires, class as a fundamental axis of social position that shapes and is shaped by race and gender. A Marxist intersectional analysis asks not only how AI affects Black women but how it affects poor Black women, migrant women of colour, women of colour in precarious employment.

Postmodern theory contributes a scepticism about stable identity categories, reminding us that “Black woman” or “working-class immigrant” are themselves constructed categories whose boundaries are contested and whose meaning shifts across contexts. This does not undermine intersectional analysis, it refines it, preventing it from reifying the very categories it seeks to use critically.

Together, the four frameworks produce an intersectional analysis of AI that attends to race, gender, class, and the epistemological dimensions of how knowledge about people is produced and used, a richer analysis than any single framework can deliver.

Reflection question: Think of an AI system that affects people across multiple dimensions of social position. What would a genuinely intersectional analysis of its impacts reveal?