Presentation Information

[4H5-OS-4c-04]The Impact of performance trust on cognitive workloadSEMによるPerformance Trustの影響の分析

〇Mari Saito1, Seiji Yamada2 (1. The Graduate University for Advanced Studies, SOKENDAI, 2. Kanagawa University)

Keywords:

Image Generation,Trust,Cognitive workload,Generative AI,SEM

Generative AI is widely used in a variety of contexts, including professional work and creative activities; however, iterative prompt trials required to obtain outputs that match users’ intentions may increase cognitive workload. This study experimentally investigated how trust in AI—particularly Performance Trust—relates to cognitive workload during prompt-based interaction with an image-generation system. Participants performed image-generation tasks under two conditions: an Automatic condition without text input and a Prompt condition requiring textual input. Trust in AI and cognitive workload were measured using subjective assessments. Structural equation modeling revealed that higher Performance Trust was associated with a significant reduction in cognitive workload, and item-level analyses further indicated differential contributions among trust components. These findings suggest that, in designing low-burden human–AI collaboration, it is necessary to consider not only the level but also the qualitative aspects of trust.