Is Auditive Communication with ChatGPT an Effective Means of Building Trust Between People and Machines: A Quantitative Study

Rapid advances in Artificial Intelligence (AI) have accelerated the integration of conversational agents into everyday tasks. While voice-based interaction is becoming increasingly prevalent, its influence on user trust in AI systems remains insufficiently understood. Existing research has largely focused on text-based interfaces, leaving open whether auditory interaction can enhance or even diminish perceived trustworthiness. This study empirically examines whether the communication modality of ChatGPT (text vs. auditory) affects users’ trust in the system. In a controlled experiment, participants with diverse backgrounds interacted with ChatGPT to complete story-based tasks requiring nuanced reasoning. Trust was measured through a nine-item quantitative questionnaire grounded in the Technology Acceptance Model (TAM). The results show that speech-based interaction was associated with significantly higher general trust in technological systems (Q1: p = 0.019, d = 0.59). No significant differences were found for perceived truthfulness, doubts about system accuracy, usefulness, or ease of use. These findings suggest that trust formation depends less on the interaction channel and more on underlying system qualities, such as accuracy, coherence, and conversational competence. The study provides new insights for designers of AI-driven voice systems: resources should be prioritised toward improving response quality and transparent system behaviour rather than assuming inherent trust benefits from auditory communication.