SocialCounterfactuals: Probing and Mitigating Intersectional Social Bias with Counterfactuals

Explore a 28-minute research presentation from the University of Central Florida that introduces SocialCounterfactuals, a novel approach for investigating and addressing intersectional social bias in AI systems. Learn how researchers use counterfactual reasoning to identify and mitigate harmful biases across multiple social dimensions simultaneously, offering more nuanced understanding of how AI models handle complex social identities. The talk demonstrates practical applications of this methodology and discusses its implications for creating more equitable and fair artificial intelligence systems.