Painting A Holistic Picture of Trust in and Adoption of Conversational Agents: A Meta-Analytic Structural Equation Modeling Approach

Abstract

With their human-like nature, conversational agents (CAs) introduce a social component to human-computer interaction. Numerous studies have previously attempted to integrate this social component by incorporating trust into models such as the technology acceptance model (TAM) to decipher the adoption mechanisms related to CAs. Given the heterogeneity of these previous works, the aim of this paper is to integrate empirical evidence on the role and influence of trust within the nomological network of the TAM. For this purpose, we conduct a meta-analytic structural equation modeling approach based on 45 studies comprising k = 155 correlations, and N = 13,786 observations. Our findings highlight the multifaceted role of trust as a mediator transmitting the effects of the technology-related perceptions that drive the intention to use CAs. Our results present a comprehensive overview in a thriving research field that can guide both future theory building and the designs of more trustworthy CAs. © 2022 IEEE Computer Society. All rights reserved.