Emotional support or emotional release? A dual-path model of emotional intelligence of AI companions on user engagement

Purpose The primary objective of this research is to explore the underlying mechanisms of AI companions' emotional intelligence influencing user engagement through a dual-path model including both perceived emotional support and perceived emotional release. Design/methodology/approach An online survey of 251 users of AI companions in China was conducted to test the proposed research model and hypotheses. The partial least squares (PLS) approach was used to analyse the data. Additionally, this study provides further insights through interviews. Findings The results show that emotional intelligence of AI companions enhances users' perceived emotional support and emotional release. Perceived emotional support is positively associated with both hope agency and hope pathways, while perceived emotional release is positively associated with hope pathways but not hope agency. Emotional release tendency moderates the relationship between AI companions' emotional intelligence and users' perceived emotional release, while emotional support tendency does not moderate perceived emotional support. Originality/value This study proposes a multidimensional framework for AI companions' emotional intelligence, based on the interpersonal emotional intelligence model. It examines how emotional support and emotional release affect user behaviour simultaneously and explores the role of hope in this mechanism. Additionally, it highlights the importance of user tendencies in shaping the impact of AI companions' emotional intelligence on user emotional perception.