Human-AI Interaction – Is It Trust or Emotions That Mediates Behavioral Intensions?

Abstract

In Human-Artificial Intelligence (AI) interaction research, trust is the dominating research domain. However, based on some recent evidence, trust feelings are not as holistic determinants as emotions in predicting eventual reactions. The objective of this research is to compare whether trust or emotions are better predictors for intentions in AI domain. Accordingly, the study adopts concepts of trust and emotions. The dual process theory offers a metalevel framework explaining some of our findings. The results imply that in contrast to the prevailing research domain and opinion, emotions play a more significant role in predicting intensions than trust. In addition, AI-generated negative images seem not to raise such emotions that would have any significant effect on intentions. Instead, AI-generated positive images created emotions that had significant effects. According to the previous literature, the negative cues emphasize the effects hence, our results challenge this view.