A Nexus of Explainability and Anthropomorphism in AI-Chatbots

Abstract

AI chatbots are transforming customer service, yet their black-box nature often undermines consumer trust and hinders adoption. Drawing on Explainable AI (XAI), Computers as Social Actors (CASA), and Trust in Automation (TiA) frameworks, this study examines how explainability and anthropomorphism shape consumer trust in AI chatbots and influence adoption across high-stakes (Finance) and low-stakes (Retail) industries. Our findings reveal that explainability strengthens consumer trust, which drives adoption, reinforcing its role in building consumer confidence. Additionally, chatbots that are both explainable and anthropomorphic generate stronger consumer trust, though the influence of anthropomorphism differs by industry. In high-stakes industries, where transparency and reliability are critical, chatbot design should emphasise clear, structured explanations while ensuring that anthropomorphic cues support rather than overshadow trust-building. Conversely, in low-stakes industries, where engagement and interaction quality may take precedence over transparency, anthropomorphic cues play a more significant role in trust formation. Despite these differences, consumer trust remains the strongest predictor of AI chatbot adoption across both industries, highlighting its fundamental role in AI acceptance. This study provides a theoretical framework and practical recommendations for designing AI chatbots that effectively balance explainability and anthropomorphism to enhance consumer trust and adoption in different industry contexts.