Banking on voice: AI attributes, technology perceptions, and trust in banking voicebot acceptance

Abstract

As Malaysian banks remain reliant on text-based chatbots, the anticipated shift toward AI-enabled voicebots highlights a technology–practice gap and the need to understand user trust and adoption in high-risk financial contexts. This study develops and tests an integrative model grounded in the Technology Acceptance Model (TAM), socio-communicative perspectives (Anthropomorphism, Social Presence, Media Richness), and a dual-dimensional trust framework distinguishing cognitive and emotional trust. A cross-sectional survey of 448 Malaysian adults, recruited via purposive sampling after viewing a banking voicebot demonstration, was analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM) across eleven latent constructs. Results show that anthropomorphism, social presence, and media richness significantly influence usefulness, enjoyment, and cognitive trust, while anthropomorphism and social presence also affect emotional trust. Emotional trust emerged as the strongest predictor of adoption intention, whereas ease of use was non-significant once trust and enjoyment were considered. The study contributes by extending TAM with AI-specific socio-communicative cues and dual trust, demonstrating that emotional trust—rather than usability—is central in high-stakes adoption, and offering practical insights for banks to prioritize conversational naturalness, social presence, and reassurance features when designing and deploying voicebots.