The Architecture of AI-Trust: A Scoping Review on Conceptualizations, Trust Bases and Antecedents

Abstract

The proliferation of Artificial Intelligence (AI) technologies, such as ChatGPT, has intensified the need to understand human–AI interaction, especially the role of trust. While prior research has explored individual aspects—like cognitive or emotional trust bases and various antecedents—it lacks an integrated perspective. This paper fills that gap through a scop-ing review of 57 peer-reviewed studies, developing a comprehensive framework of trust in AI. The frame-work categorizes trust antecedents into user-based, context-based, and dynamic factors, and identifies five key trust bases: cognitive, emotional, moral, institu-tional, and transitive. It also distinguishes between AI-trust conceptualizations, where AI is seen as human-like, technological, or provider-based. By unifying these elements, the study provides a holistic view of AI trust, identifies research gaps, and offers guidance for future studies to enhance both initial and ongoing user trust in AI systems.