Psychological Traits and Appropriate Reliance: Factors Shaping Trust in AI

Abstract

Research in AI-enabled decision support often focuses on technological factors influencing reliance on AI. However, the end-users of AI systems are individuals with diverse personalities which potentially lead to differences in collaborative human-computer interaction, resulting in harmful under- and over-reliance. The influence of psychological traits on appropriate reliance must be understood to enable development of more effective AI support addressing a diverse user base. This experimental mixed-methods study (N = 250) examined the impact of psychological traits on trust in and reliance on AI advice in classification tasks. Propensity to trust, affinity for technology interaction, and control beliefs in interacting with technology were identified as predictors for trust, which affect reliance. Thus, consideration must be given to the expected propensity to trust and the level of technological expertise among user groups when designing systems that aim to promote suitable degrees of trust and appropriate reliance.