Acceptance Barriers and Interaction Mechanisms of Communication- Apprehensive Users in Generative AI Voice Conversations

Previous research has identified a ‘Safe Haven’ effect in which users with high communication apprehension perceive text-based AI chatbots as psychologically safe conversational spaces. This study examined whether this effect persists in generative AI voice user interface (VUI) environments characterized by real-time and irreversible interaction. An explanatory sequential mixed-methods approach was employed, including a survey of 108 users and in-depth interviews with five users exhibiting high communication apprehension (CA). Quantitative results showed no significant CA-based differences in usage motivation, social presence, or interaction behavior, while perceived ease of use (PEOU) was significantly lower among high-CA users. Qualitative analysis revealed psychological and cognitive bur- dens related to the irreversibility of voice input and mechanically paced turn-taking. These findings suggest that the safe zone mechanism identified in text-based interaction does not operate identically in voice environments.