Trust Differentiation in Human-Computer Interaction: The Dual-Path Impact of AI Voice Assistant Role Types on Continuance and the Moderation of Self-Construal

Abstract

Despite the rapid adoption of AI voice assistants, user retention remains generally low. Investigating how to enhance user trust and continuance intention is therefore of significant theoretical and practical importance. Grounded in role theory and the Stimulus-Organism-Response (SOR) framework, this study categorizes AI voice assistants into two role types: companion-type (emotion-oriented) and butler-type(task-oriented). We constructed a theoretical model positing: "role type ? affective/cognitive trust ?continuance intention " and introduced Self-construal as a moderator. Data collected via scenario-based experiments were analyzed using SPSS for regression and Bootstrap mediation tests. This approach systematically examines the differentiated mechanisms through which AI voice assistant roles influence trust and continuance intention. The results reveal that companion-type assistants significantly foster continuance intention by strengthening users' emotional trust. In contrast, while butler-type assistants establish cognitive trust, they suppress continuance intention by triggering privacy concerns and perceived loss of control. Furthermore, users with an interdependent self-construal exhibit stronger emotional trust and higher continuance intention toward companion-type assistants. However, an independent self-construal does not significant moderate the effect of butler-type assistants.

Keywords: AI voice assistant, continuance intention, companion-type assistant, butler-type assistant, trust, self-construal