“There’s no such thing as a single-issue struggle, because we don’t lead single issue lives”- Audre Lorde 

Introduction to Intersectionality

Intersectionality: a term coined by Kimberlé Crenshaw in 1989 to capture what it truly means to lead unique lives. It originates in the inability to capture the specific nature of the discrimination (due to the overlap of race and gender) faced by a Black woman named Emma DeGraffenreid at her workplace. The case inspired Crenshaw to search for a lens to name the problem so DeGraffenreid’s experience could at least be acknowledged. She used the idea of intersections to aid in visualising and developing a frame which could highlight the woman’s experience.  

Intersectionality can be understood as a framework through which one can understand the unique combination of privileges and discrimination that a person faces due to the overlap of various categories of their identity, including (but not limited to) race, gender and class. 

Every being on this planet occupies a specific position, the terms of which are dictated by aspects both in and out of our control. The elements that work in your favour to uplift your circumstances are called ‘privileges’. The elements that work against you to create a discriminatory experience are a result of existing ‘structures of oppression’. The overlap of our privileges and discrimination forms our unique experiences and struggles. Oftentimes there is a possibility of not (actively) realising the influence of oppressive structures due to our existing privileges, resulting in what is called a ‘privilege hazard’.  

There is an underlying theme among all the above, which pulls the strings: power. Our experiences are dictated by this puppeteer’s actions as it determines how it manifests and distributes itself within our lives. The impact of power extends from pre-determining the possibility of certain experiences within one’s life to influencing socio-technical definitions and understanding of fairness, representation and bias within systems that dictate the shaping of society.  

Introduction to AI Fairness

AI fairness is a specialisation within AI research aimed at understanding, exposing and reducing the risks and harm caused due to bias in AI systems. However much of the research approaches fairness as a problem that is fixable with a technical band-aid. This tends to skew the notion of fairness to either statistical parity (minimising negative outcomes equally among different groups) or intersectional subgroup fairness. The issue with the first notion of fairness is that while this appeals to the idea that there should be no discrimination against any kind of identity group, it also assumes that all identity groups have the same starting line. Furthermore, it does not account for the unique experiences caused by overlapping identities. While the latter notion does indeed take into consideration different sub-groups which constitute a person’s identity, it runs into the problem of not being able to account for every sub-group. Thus, both these definitions of fairness are incomplete.  

Our Call-to-Action

To truly reach a future which is just, it requires fairness to be supported by equity and equality. It begins with acknowledging that while we have made significant leaps and bounds in fairness research, there is always more that we can do. When we challenge our assumptions, it pushes us to look beyond what we want to see in our data and educate ourselves on what the data is reflecting. When we share our knowledge and responsibility of mitigating bias issues, we have the potential to create multidimensional solutions. In the following column, we explore the different elements related to intersectional AI fairness and look at some action-oriented recommendations to apply intersectionality to AI fairness. In doing so, we can call upon a future where justice is the norm, collaboration is key and social change is a priority. 

It takes one conscious moment to make a choice; join us in making one that will sustain generations. 

Next Article ->