AI and Racism are both controversial and sensitive topics. But now it’s time to peel back the covers and take a peek at how they can be interrelated.
Racism can lead to discrimination in various areas of life, such as employment, housing, and education. This can limit access to opportunities and resources for people of color, making it more difficult for them to succeed and improve their lives. Relating to mental illness, racism exposes one to psychological harm as well. Experiencing racism can cause stress, anxiety, and depression, and can lead to a sense of hopelessness and powerlessness as well as damage self-esteem and self-worth.
Racism has been causing a public disturbance as well as social and economic inequality for quite a while now. Racism leads to disparities in income, wealth, and access to resources, which can perpetuate cycles of poverty and disadvantage for People of Color. Since racism is baked into our society, and our society is made up of people – and people design, develop and interact with technology, it is not a far stretch to conclude that technology has the propensity to perpetuate racism. Just because you don’t know about it – doesn’t mean that it’s not happening.
Artificial Intelligence and Algorithmic Black Box
First things first, what’s AI racism? AI (Artificial Intelligence) racism refers to the potential for AI systems to perpetuate or exacerbate racial biases. AI systems are trained on data, and if the data used to train them is biased, the AI systems can develop and reflect that bias. This can lead to discriminatory outcomes in areas such as hiring, lending, advertising, and criminal justice. Penned by Calvin D. Lawrence, Hidden in White Sight: How AI Empowers and Deepens Systemic Racism is a book about how AI isn’t functioning the way it was supposed to and instead, is reflecting prejudice.
One example of AI racism is facial recognition technology. Studies have shown that these systems are less accurate in identifying people with darker skin tones and are more likely to misidentify them. This can have serious consequences, such as wrongful arrests and false accusations. Another instance is predictive policing, where AI systems are used to predict crime and identify potential suspects. If the data used to train these systems is biased, it can lead to the over-policing of certain neighborhoods, disproportionately affecting communities of color.
Artificial Intelligence for Humanity – Friend or Foe?
Alongside various areas of concern, AI can also perpetuate bias in the hiring process. If AI systems are trained on resumes submitted in the past and these resumes are mostly from privileged groups, the AI systems will identify similar qualifications in future resumes as well, leading to a lack of diversity in hiring. It’s important to recognize the potential for AI systems to perpetuate racial biases and to work towards developing AI systems that are fair and unbiased.
This can be done by collecting and analyzing diverse data sets, involving people from marginalized communities in the development and testing of AI systems, and regularly evaluating and monitoring the performance of AI systems to ensure that they are not producing discriminatory outcomes. However, it’s important to note that AI itself is not inherently racist, but it can perpetuate and amplify the bias that exists in the data it was trained on.
Can AI-Based Racism Be Cured?
Well, that’s a trick question. Asking whether AI-based racism can be cured is like asking whether or not the great societal ill of racism can be eradicated. I’m not smart enough to answer that one, but I will say that it can be mitigated with proper attention and focus. Minimizing the harm caused by AI racism can be achieved by erecting proper guardrails and aligning best practices in the development and deployment of AI systems, such as:
- Ensuring that the data used to train the AI system is diverse and representative of the population it will be used with.
- Regularly monitoring and testing the AI system to detect and address potential biases.
- Incorporating transparency and explainability into the AI system to allow for easy identification of potential biases.
- Using adversarial training and counterfactual analysis techniques to reduce the impact of biases in the AI system.
- Involving diverse stakeholders in the design and deployment of the AI system to ensure that the perspectives of different groups are considered.
