Explainable AI in Recommendation Systems: Enhancing Transparency and Consumer Trust in Business Analytics

Abstract

A midst the age of data-driven business, systems of recommendation have taken center stage in online business strategies, as they customize consumer experiences and increase interactivity. Nevertheless, the lack of transparency, accountability, and consumer trust in most artificial intelligence (AI) models has been an issue due to the black-box nature of the models. The paper examines the role of Explainable artificial intelligence (XAI) in making recommendation systems interpretable and acceptable in business analytics, with reference to the E-ReDial dataset, which includes explainable conversational recommendation data. The study makes a conceptualization that connects explainability, transparency, consumer trust, and adoption intention in AI-driven business systems. An explanation-generation layer was applied to a baseline recommender model that was trained on collaborative filtering so as to give textual rationales to recommendations. Empirical results show that the quality of explanation has a significant positive impact on perceived transparency (0.63, p <.001) and trust (0.54, p <.001), and transparency is partially mediating the relationship between explainability and trust. Moreover, a stronger trust level has a positive impact on recommendation adoption, which can be considered an important business impact of explainable AI integration. The analysis points out that the explanation fidelity, clarity, and contextual relevance are some of the most important factors that allow determining consumer confidence in AI-mediated decisions. In managerial perspective, the integration of explainability in business analytics pipelines may enhance user interaction, conversion, and ethicality through accountability of the models. The study has a theoretical impact in that it empirically confirms the connection between clarification and trust and a practical contribution of the study in the formulation of a framework of transparent and human-friendly AI systems in digital business markets. This structure could be further developed into multi-domain conversational recommenders in the future, and longitudinal trust interactions in real-world applications could be explored.

Keywords: Explainable Artificial Intelligence (XAI), Recommendation Systems, Business Analytics, Consumer Trust, Transparency, E-ReDial Dataset

View PDF