Could AI Be the Catalyst for a Racially Just Society?

As we venture further into the digital age, a pivotal question arises: Could the rapid advancement of Artificial Intelligence (AI) help create a more racially equitable society? Although it might seem paradoxical given AI’s history of reinforcing racial biases, emerging AI applications could provide novel ways to address these very issues.

Racial disparities are deeply embedded in various sectors—healthcare, law enforcement, and economic opportunities, to name a few. Despite this, society often focuses more on correcting biases in AI systems than on grappling with those in human behavior. This unusual focus presents a unique opportunity to use AI as a tool to reflect, quantify, and ultimately mitigate societal biases.

AI technology comes with three main advantages that could be utilized in promoting racial equity. First, AI can highlight existing biases through the historical data it processes. By making these biases visible, AI challenges us to confront uncomfortable truths within our institutions. Second, unlike humans, AI systems can be systematically tested, evaluated, and improved. Techniques like metric elicitation allow stakeholders to define fairness within specific contexts, helping to develop solutions that address historical inequities without resorting to superficial “colorblind” approaches. Third, AI can reduce implicit biases in decision-making processes by relying on explicitly programmed relevant factors, thereby avoiding the unconscious biases that can color human decision-making.

To harness AI’s potential, we must address both technical and social dimensions. Technically, deploying sophisticated approaches is critical, especially where biases persist in high-stakes fields like healthcare and criminal justice. Socially, the development and deployment of equitable AI systems require inclusive processes that incorporate diverse voices. Initiatives by organizations such as Black in AI emphasize the importance of involving impacted communities to ensure that AI systems operate fairly.

The path forward involves using AI not as a replacement for human judgment but as an augmentative tool for ensuring equity. By drawing attention to biases, providing frameworks for refinement, and fostering explicit fairness considerations, AI can play a significant role in building a racially just society—a goal that remains out of reach through human efforts alone. AI’s potential lies in its ability to prompt us to reflect on our values and decisions, urging us to actively seize this moment to promote justice, not perpetuate harm.

Sanmi Koyejo, an assistant professor of computer science at Stanford University and the leader of the Stanford Trustworthy AI Research team, emphasizes the need for deliberate engagement with AI technology, ensuring that its impact on racial equity is as beneficial as possible for society as a whole.