Cultural Bias in Explainable AI Research: A Systematic Analysis

Western bias dominates XAI research: The paper's systematic review of over 200 XAI user studies found that most tested only Western populations while drawing broad conclusions about how humans generally interact with AI explanations, despite psychological evidence showing significant differences in explanatory preferences between Western individualist and non-Western collectivist cultures.

Current XAI designs may not work for everyone: Popular explainable AI systems are built on assumptions about what makes explanations helpful that reflect Western, individualist values—such as focusing on individual features or personal agency—but these designs may be ineffective or even confusing for people from collectivist cultures who prioritize group context and relational information.

A critical knowledge gap exists: The paper identifies that XAI researchers have largely overlooked how cultural background shapes what people need from AI explanations, leaving a significant blind spot in our understanding of whether widely deployed XAI systems actually serve diverse global populations effectively.

Literature reviews perpetuate the problem: Analysis of over 30 XAI literature reviews revealed that most failed to acknowledge cultural differences in explanatory needs or caution against overgeneralizing findings from Western-only studies, allowing biased assumptions to persist and spread throughout the field.

Urgent action needed for inclusive AI: The findings suggest that developing truly responsible AI requires XAI researchers to actively test their systems with culturally diverse user groups and incorporate cultural considerations into both system design and research methodology, rather than assuming one-size-fits-all explanations.