Influence of Explainability and Presentation Modality on User Cognition, Trust, and Decision Making in an Intelligent Fire Evacuation System

The effectiveness of an intelligent fire evacuation system (IFES) depends on human−AI interaction. This study evaluated the influence of explainability and presentation modality on cognition, trust, and decision-making in fire evacuation. Results revealed that while high explainability significantly enhanced trust by 30% and confidence by 34% relative to low explainability, these benefits were accompanied by an 82% increase in response latency after recommendation onset, which partly reflected the longer time required to process more detailed recommendations. Conversely, medium explainability emerged as the optimal design, balancing the need for psychological reassurance with the imperative for rapid action. Regarding presentation modality, visual modality facilitated the shortest response latency through efficient scanning, reducing decision time by 49% compared to the auditory modality, while audiovisual modality was most effective in fostering trust and satisfaction. Consequently, this study advocated for the development of an adaptive IFES that optimizes rather than maximizes information, flexibly deploying modalities.