Penn State Researcher Proposes Socioculturally Competent AI to Combat Racial Bias

Christopher Dancy, a Penn State associate professor, presented a critical analysis of the intersection between artificial intelligence and racial identity on September 10. The lecture, titled "Critically Considering the Human and AI in Human-AI Interaction," argues that AI systems are not neutral but actively reflect and shape concepts of identity and Blackness.

Speaking as part of the IST Research Talks series, Dancy detailed a multi-level framework designed to scrutinize AI engineering processes. His research specifically examines how these computational systems impact the preservation of Black digital archives and the efficacy of disaster recovery efforts. Dancy holds joint appointments across Penn State's Colleges of Information Sciences and Technology (IST), Engineering, and the Liberal Arts, bringing a cross-disciplinary lens to the technical failures of current AI models.

The Framework for Competence

The research is underpinned by a U.S. National Science Foundation (NSF) CAREER Award, which supports Dancy's objective of creating "socioculturally competent" AI systems. This approach moves beyond simple bias mitigation to address the foundational ways racialization and anti-Blackness are embedded into the architecture of AI. Dancy's work is further supported by his leadership roles as the founder and director of the Human in Computing and Cognition Lab and the Liberatory Tech Project, as well as his role as co-director of the Center for Black Digital Research/#DigBlk (CBDR).

Why Sociocultural Competence Matters

This research addresses a critical gap in AI engineering: the tendency for high-stakes systems to fail marginalized communities due to a lack of cultural context. In the context of disaster recovery, for example, AI that lacks sociocultural competence may overlook specific community needs or misinterpret data from Black neighborhoods, leading to inequitable resource distribution. Similarly, in digital archiving, biased AI can erase or miscategorize historical records essential to Black identity.

By analyzing the engineering pipeline, Dancy seeks to ensure that the tools used to manage human data do not perpetuate systemic exclusions. The goal is to transition from AI that merely processes data to systems that understand the social and historical contexts of the people they serve.

The Path Forward

The presentation is part of a broader initiative at Penn State to prioritize human-centered and ethical AI innovation. As the university continues to showcase interdisciplinary work through the IST Research Talks, the focus remains on how academic frameworks can be translated into engineering standards. Future developments will likely center on whether these socioculturally competent models can be scaled across other high-stakes public sectors to prevent the automation of racial bias.