AI transparency and bias mitigation in customer service: how ethical culture shapes customer trust

Abstract

This study aims to examine an unresolved issue in artificial intelligence (AI) driven customer service: Why transparency and bias mitigation do not automatically translate into customer trust across organizational contexts. Specifically, this study investigates the effects of bias mitigation strategies and AI transparency on customer trust, and the moderating role of organizational ethical culture, in the context of China. A cross-sectional survey of 1,450 customers using generative AI services in banking and e-commerce was conducted. Adapted scales measured perceptions of AI transparency, bias mitigation, customer trust and organizational ethical culture. Data were analyzed using Partial Least Squares Structural Equation Modeling via a two-step approach: assessment of the measurement model (reliability and validity), followed by the structural model to test hypothesized relationships. Multiple regression served as a supplementary robustness check. The study found that both bias mitigation strategies (β = 0.30, p < 0.001) and AI transparency (β = 0.35, p < 0.001) significantly enhance customer trust. Additionally, organizational ethical culture moderated these effects, amplifying them (β = 0.20 and 0.18, p < 0.01). These results emphasize the importance of transparency and fairness in AI, particularly in organizations with strong ethical cultures. This study reframes AI transparency and bias mitigation as governance and organizational control mechanisms, rather than mere technical features. It demonstrates that organizational ethical culture acts as a key informal context, shaping whether customers view these mechanisms as credible signals of fairness and accountability. In the Chinese setting, the study explains why transparency and bias mitigation do not automatically build customer trust.