Paper Type
Short
Paper Number
PACIS2026-1926
Description
As generative AI increasingly functions as a social partner, understanding its psychological effects on users has become important. Drawing on the Machine-Integrated Relational Adaptation (MIRA) model and Self-Expansion Theory, this study examines how AI voice modality (neutral text, machine-like voice and human-like voice) influences users’ social confidence through perceived intelligence and self-expansion, with attitudes toward AI serving as a moderator. A 14-day longitudinal experiment with 125 university students was conducted in which participants interacted daily with an AI companion under different modality conditions. Preliminary results show that human-like voice significantly increases perceived intelligence; although direct effects on self-expansion and social confidence are not observed in simple group comparisons, the findings suggest indirect psychological mechanisms linking AI voice design to users’ social confidence. This study extends to Human-AI interaction and social development literature and provides implications for designing AI companions that support users’ psychological empowerment and social development.
Recommended Citation
Liu, Yang and Wei, Liuyan, "From Voice to Valor: An Examination of How AI Voice Modality Enhances User Social Confidence" (2026). PACIS 2026 Proceedings. 12.
https://aisel.aisnet.org/pacis2026/hci_robotic/hci_robotic/12
From Voice to Valor: An Examination of How AI Voice Modality Enhances User Social Confidence
As generative AI increasingly functions as a social partner, understanding its psychological effects on users has become important. Drawing on the Machine-Integrated Relational Adaptation (MIRA) model and Self-Expansion Theory, this study examines how AI voice modality (neutral text, machine-like voice and human-like voice) influences users’ social confidence through perceived intelligence and self-expansion, with attitudes toward AI serving as a moderator. A 14-day longitudinal experiment with 125 university students was conducted in which participants interacted daily with an AI companion under different modality conditions. Preliminary results show that human-like voice significantly increases perceived intelligence; although direct effects on self-expansion and social confidence are not observed in simple group comparisons, the findings suggest indirect psychological mechanisms linking AI voice design to users’ social confidence. This study extends to Human-AI interaction and social development literature and provides implications for designing AI companions that support users’ psychological empowerment and social development.
Comments
12-HCI