The Influences of AI Locus of Causality and Trust in AI on the Adoption of AI Advice

This study aims to delve into a mechanism explaining how individuals’ perceptions of AI responsibility are related to their adoption of AI advice based on attribution theory. We propose a concept of AI locus of causality (LOC-AI), which represents an individual’s perception of the extent to which AI influences decision-making performance. We built AI-embedded websites for a longitudinal experiment in which participants made decisions on corporate credit ratings and used a panel dataset collected from the experiments. The longitudinal evidence showed that individuals’ decision performance influenced trust in AI but not LOC-AI. The results also indicated a positive lagged effect of LOC-AI on the adoption of AI advice but not trust in AI. Overall, we found that individuals were likely to exhibit self-serving biases and to adopt an egocentric and disengagement coping strategy even though they tended to leverage AI advice in their decisions.