What is intersectionality?

Intersectionality, as defined by Kimberlé Crenshaw, is a framework for understanding how overlapping forms of inequality such as racism, sexism, and classism interact to create unique barriers for certain groups. When we apply this lens to Artificial Intelligence and young women, particularly those from marginalised racial, ethnic, and socio-economic backgrounds, the risks of compounded discrimination become clear.

Intersectional bias in AI

Real world consequences

The limited data availability of intersection data, as discussed by Alan Turing Institute, compounds the problem. Much AI research and policy still treats gender as binary and fails to examine race or other intersecting identities in meaningful depth. This narrow focus, often centering “women in tech” in ways that privileges white women, overlooks the specific barriers faced by LGBTQ+ women, BAME women, disabled women, and those at other intersections of marginalisation. The danger here is twofold, as AI both inherits bias from unequal social systems and amplifies it through automated decision-making at scale.

For instance, ProPublica’s 2016 investigation into the COMPAS risk assessment tool revealed that Black defendants were nearly twice as likely as white defendants to be incorrectly labelled high risk. Similarly, Lum and Isaac’s research on the PredPol predictive policing system demonstrated how biased policing data could create feedback loops that further target Black and Brown neighbourhoods, not because crime rates are higher, but because biased reporting and enforcement make those communities appear more criminal in the dataset.

Amid these systemic issues, campaigns like Exposed, led by Amina, the Muslim Women’s Resource Centre, are crucial. Exposed highlights a form of abuse that is often overlooked: sharing someone’s private photos without their consent. For many Muslim and Black and Minority Ethnic women, even images that are not sexual, like a photo without Hijab or with uncovered arms, can cause serious harm. These pictures can lead to shame, family pressure, honour-based abuse, or even forced marriage. Right now, Scottish law only protects people if the images are sexual, leaving many women with no legal protection.  Exposed helps raise awareness and deepen understanding of the real impact this abuse can have, especially within cultural and religious contexts that are often overlooked in broader discussions of image-based abuse.

Looking to the future

While much of the evidence highlights harm, there are glimpses of AI’s potential for good. UNESCO (2020) points to tools that can help employers write gender-sensitive job postings to attract a more diverse workforce, illustrating that AI can support equality when designed intentionally and inclusively. But such outcomes require an explicit commitment to intersectionality in both policy and practice. 

As Collect and Dillon (2019) argue, this means drawing on feminist, queer, postcolonial, and critical race theory to understand how AI interacts with existing power structures and to ensure law and policy address not just individual inequalities but their intersections. For young women, especially those who live within the margins, intersectional approaches to AI are essential for building systems that do not silently replicate the very inequalities they promise to solve. Without them, the technology of the future risks becoming the oppression of the present, only faster and more efficient.