Illuminating the AI Black Box: The Impact of Explainable AI on Perceived Justice and Fairness

Chuan, C. H., & Ruoyu Sun. “Illuminating the AI Black Box: The Impact of Explainable AI on Perceived Justice and Fairness,” paper to be presented at the International Communication Association (ICA) Annual Conference, Denver, June.

Abstract: As AI becomes increasingly integral to high-stakes decision-making, such as loan approvals and criminal justice, eXplainable AI (XAI) is vital for enhancing the transparency of these systems. This study investigates the impact of different XAI types on users’ perceptions of justice and fairness, focusing specifically on global and local XAI approaches. The online experiment included three well-known XAI explanations: a global XAI explanation using a transparent model (decision trees) and two local XAI explanations (example-based and counterfactual explanations). A one-factor experimental design was conducted with 145 participants randomly assigned to one of four conditions, including a control condition without any explanation. The findings indicate that XAI can effectively improve perceptions of procedural and informational justice and algorithmic fairness, and the effectiveness varies by the type of XAI. Directions for future research on XAI in communication are provided.