Voice Design and Trust in Automated Vehicles: Findings and a Research Agenda
作者
Jiongyu Chen,Qiaoning Zhang
出处
期刊:日期:2025-11-12卷期号:: 688-696被引量:1
标识
DOI:10.1109/tps-isa67132.2025.00085
摘要
Most work on human-automated vehicle (AV) interaction has asked what an AV should explain rather than how it should sound while explaining. Building on this gap, we test how voice gender, user-voice gender similarity, and gender-role congruity shape cognitive and affective trust in AVs. In an online experiment with 326 U.S. drivers who viewed six driving scenarios with “what + why” explanations delivered by synthetic male or female voices, we found that gender similarity increased affective trust but did not change cognitive trust; this affective gain appeared only when the voice fit the role participants perceived the AVs to occupy, and it disappeared under role incongruity. Interpreted through Computers Are Social Actors (CASA), similarity attraction, and role congruity theory, the results identify distinct pathways for trust formation: emotional comfort is sensitive to social alignment and role expectations, whereas competence judgments rely on informational quality. We propose a research agenda that separates cognitive and affective trust, advances fairness-aware voice design beyond binary gender through non-gendered and androgynous options, and specifies context-adaptive communication that adjust tone, framing, and explanation depth to situational demands while preserving user agency. Methodologically, we recommend using validated instruments to measure both facets of trust, incorporating behavioral metrics that capture calibration and safety relevance, developing reproducible voice datasets with detailed parametric metadata, and conducting pre-registered, multi-site studies. The goal is to develop AV voices that earn trust appropriately by clarifying reasoning when needed, offering social reassurance when helpful, and remaining inclusive across users and contexts.