[4T1-OS-6c] OS-6

[4T1-OS-6c-04] Predicting Over/Under Trust Dynamics via Dynamic Structural Equation Modelling

〇Sota Kaneko^1,2, Seiji Yamada^2,1 (1. The Graduate University for Advanced Studies, SOKENDAI, 2. National Institute of Informatics)

Keywords:Human-Agent Interaction, Trust Dynamics, Over/Under Trust

With the rise of Artificial Intelligence (AI) technology, interest in AI-based systems like autonomous driving and chatbots is growing. Over-trust in those systems could lead to misuse, and under-trust could cause a decrease in system efficiency. In human-AI collaborative decision-making systems, such as those exemplified by autonomous driving, it is necessary to keep appropriate trust in AI to prevent system misuse and efficiency degradation. Therefore, we have constructed a predictive model of trust in AI, which dynamically changes over time, using Dynamic-Structural Equation Modeling (Dynamic-SEM), which can handle latent variables that cannot be directly observed. The constructed predictive model has been confirmed to be able to predict over-trust.