The role of Explainable AI in the Design of Visual Texts for Trust Calibration in Level 3 Automated Vehicles
Diah Ayu Irawati, Elif Bölükbaşı, Michael A. Gerber, Andreas Riener · 2024
Considerable research has been carried out into explainable artificial intelligence (XAI) in automated driving to enhance user trust in these technologies. This work examines how XAI and HCI can improve user trust through visual text design in L3 automated vehicles. We aim to develop user interfaces (UIs) that make AI-driven vehicle decisions more transparent and understandable. Employing a mixed-methods approach (TiA scale, qualitative feedback) in a between-groups design (n=12 participants), we combine empirical analyses with UX design principles. The study indicates that effectively communicating AI processes through visual texts in UIs can bridge the gap between complex algorithms and user understanding, thus fostering trust. Results further indicate that contextual, clear, and expected multimodal interaction enhances user trust and understanding of automated systems. These findings are crucial for developing future automated systems in a user-centered manner.