AI, Intersectionality and Power

In recent times, more attention is being given to the stories of those who do not fit the ‘default’ and as a result face discrimination. In today’s techno-social world, AI is one of the media through which people are reminded of their otherness. From smart home devices to digital assistants and chatbots, AI is deeply ingrained into our society. However, its integration comes with its own set of issues, ranging from over-reliance on AI models to perpetuating human bias and discrimination. 

One of the most well-known examples of AI injustice is that of Joy Buolamwini, founder of the Algorithmic Justice League. Buolamwini was a Black graduate student at MIT who found that facial recognition systems struggled to detect her features, unlike those of her lighter-skinned peers, unless she wore a white mask upon which the software had no trouble recognising the mask’s features perfectly.  

In their experience with bioimpedance-based body composition analysis tools, Kendra Albert narrates the feeling of exclusion in technology-based personal healthcare systems which often do not include trans and non-binary people.  

Robert Julian-Borchak Williams was wrongfully detained by the Detroit Police Department for 30 hours. The reason for his arrest? A false match determined by a faulty facial recognition algorithm between his driver’s license image and a photo of a wanted criminal.  

The life of a person who deviates from the ‘norm’ means either forced conformity to society standards designed for those with power or facing the consequences of being different. When diving a little deeper, we examine who sets these standards, how they come about and why they exist in the first place. On closer inspection we notice an ideal that is firmly rooted and encircling society: power. 

Understanding power

Catherine D’Ignazio and Lauren F. Klein (authors of the book Data Feminism) define power as  “the current configuration of structural privilege and structural oppression, in which some groups experience unearned advantages—because various systems have been designed by people like them and work for people like them—and other groups experience systematic disadvantages—because those same systems were not designed by them or with people like them in mind”. In other words, power is the ability to create or change the likelihood of a particular event in a person’s life due to the systems placed by society. Systems filled with decisions that are influenced by multiple existing structures of oppression and privilege. There are multiple ways in which power is upheld and experienced. It is either institutionalised through policies, laws and law enforcement, circulated through ideas portrayed in culture and media or individually experienced via personal encounters with power and oppression. All of these systems work together to keep power circulating among the dominant groups. 

There is a truth about power which goes beyond academic and cultural definitions. Power means deciding who is worthy of representation and who is not. Power means dictating a person’s character solely on stereotypes based on factors like their skin colour or where they are from. Power means determining which members of society get to be acknowledged for their existence. It means creating (actively or passively) a world where Joy Buolamwini and other dark-skinned people do not get recognised by facial recognition systems. It means branding Robert Julian-Borchak Williams and other victims of algorithmic injustice as criminals based on a false charge due to an algorithmic mix-up. It means reminding Kendra Albert and other trans and non-binary people that they do not fit the so-called “default.”  

There is no denying the chokehold that power has on our experiences. Power is deeply entrenched into every aspect of our human lives. And it has the ability to rob people of their dignity. 

Privilege and the power of knowledge

It is, however, not all gloom and doom. There are a few avenues through which we can change the way power exerts itself.  

To do that, let us understand privilege. Privilege is not inherently a bad thing; in fact, it can be quite powerful when one decides to use it to uplift people. However, privilege can also be harmful if one uses it to actively ignore reality.  

The first step to using privilege for good is acknowledging that we have it. It is important to understand the type of privilege we have and the impact it has on our life. This makes it easier to assess what your privilege can do for others, e.g., using privilege to shine a spotlight on those that are traditionally not centered. Secondly it is important to educate ourselves on concepts which secure versus challenge power. Concepts which secure power are ones which locate the problem in individuals or technical systems (e.g., fairness), whereas concepts which challenge power acknowledge power differences and work towards dismantling them (e.g., justice). When we understand how these concepts influence the systems that dictate our lives, we can begin to examine what is necessary to create an equitable future. When we shift our focus towards challenging power, we can do right by the intention of intersectionality and pursue social justice.  

We have power, lets use it for good

Oftentimes, we tend to lose sight of who the AI systems are truly for. It is easy to get swept away into the thrill of buzzwords like “innovation” or “cutting–edge solutions”. It is crucial to remember that AI does not exist in a vacuum. It is an entity fed by data gathered from society which in turn has an impact on all its members, not just the ones with power. It begins by acknowledging that we, as people in AI, are shaping techno-social society, which in turn shapes regular society. We need to critically examine the effect of the intersections we occupy to understand our positionality and document how that affects the models we make. Good intentions are not enough if technology does not also protect the people it claims to help. It is important to question if the solutions we make are sustainable and accept that incorporating a new way of thinking is necessary if it means ensuring that we aim for justice and not only fairness. Therefore, it is our active responsibility to ensure that the systems we create are built with the idea of securing dignity for all. When we place social justice at the forefront, we can strive towards building a collective meaningful existence.  

*All sources used in the article are linked below 

Actionable Recommendations

• Reflect on and discuss your own position in terms of power and privilege. It can be useful to document this when working on AI projects to see how your position influences the perspectives and decisions you make.  
• It is also helpful to document the perspectives and decisions which were made and considered throughout the lifecycle of your AI product. This promotes open communication and transparency.  
• Thoroughly examine the role your AI technology will play and ground it in societal context to ensure a realistic representation and understanding of its impact. It is also useful to redefine concepts you aim to achieve (e.g., fairness, transparency, accountability) with power and social context in mind. This allows you to align your priorities and remind yourself of the greater purpose of your AI product.  
• Acknowledge that the AI product has the potential to do some serious harm and be open to criticisms and concerns from impacted communities. Create a platform through which they can safely voice their concerns, invite them to be a part of the participatory process and make it financially viable.