Paper Type
Complete
Paper Number
PACIS2025-1365
Description
Traditionally, driver training and assessments have heavily relied on human instructors to guide learners in defensive driving. However, the increasing adoption of artificial intelligence (AI) in education presents an unique opportunity to address limitations in human resources for consistent driver training. This study extends the conversation beyond basic AI capabilities, focusing instead on the effects of gendered synthetic voices and audio fidelity in AI-driven instruction. Prior studies have investigated issues concerning gender and racial biases inherent in AI recommendations and analyses. A critical underlying cause of these challenges is that AI is engineered to emulate the human decision-making processes. By investigating how AI instructors’ voice characteristics impact immediate driving responses and subsequent performance, this study offers a new perspective on human-AI interaction. Our findings illuminate the cognitive and emotional mechanisms involved in AI-mediated instruction, suggesting that these biases could be leveraged to enhance the learning experience in driver training.
Recommended Citation
Lunina, Yulia; Choi, Ben; and Pee, Loo Geok, "The Magical Power of AI Voice on Human Cognition and Decision-Making: A Driving-Simulation Experiment with Eye-Tracking and Facial Expression Analytics" (2025). PACIS 2025 Proceedings. 7.
https://aisel.aisnet.org/pacis2025/conftrack/conftrack/7
The Magical Power of AI Voice on Human Cognition and Decision-Making: A Driving-Simulation Experiment with Eye-Tracking and Facial Expression Analytics
Traditionally, driver training and assessments have heavily relied on human instructors to guide learners in defensive driving. However, the increasing adoption of artificial intelligence (AI) in education presents an unique opportunity to address limitations in human resources for consistent driver training. This study extends the conversation beyond basic AI capabilities, focusing instead on the effects of gendered synthetic voices and audio fidelity in AI-driven instruction. Prior studies have investigated issues concerning gender and racial biases inherent in AI recommendations and analyses. A critical underlying cause of these challenges is that AI is engineered to emulate the human decision-making processes. By investigating how AI instructors’ voice characteristics impact immediate driving responses and subsequent performance, this study offers a new perspective on human-AI interaction. Our findings illuminate the cognitive and emotional mechanisms involved in AI-mediated instruction, suggesting that these biases could be leveraged to enhance the learning experience in driver training.
Comments
Diversity