Abstract Background The quality of therapeutic relationships is a core indicator for predicting the effectiveness of psychological interventions. Traditional automated psychological education systems are mostly based on logical text feedback, and due to the lack of emotional resonance, it is often difficult to establish deep trust with users. Emotion computing technology endows machines with the ability to recognize and express emotions. The research aims to explore the impact of artificial intelligence (AI) emotion computing technology on the construction of therapeutic relationships in mental illness education scenarios. Methods A total of 240 college students with mild psychological distress were randomly assigned to two groups: (1) the Affective Computing AI group (n = 120), which received facial expression recognition and emotional feedback through voice tone; (2) the Traditional AI group (n = 120), which only received standardized text-based logical feedback. Participants completed a 4-week mental health education course. The therapeutic alliance was assessed using the Working Alliance Inventory (WAI), and course completion rates were recorded. Results: No significant differences were observed in baseline indicators between the two groups. After intervention, the affective computing AI group showed significantly better performance in therapeutic relationship building compared to the traditional AI group. The post-intervention comparison of therapeutic relationship and compliance indicators between the two groups is presented in Table 1. As shown in Table 1, the WAI total score of the affective computing AI group was significantly higher than that of the traditional AI group (p.001, d = 0.65). In the sub-dimensional analysis, the most significant difference was observed in the Emotional Bond dimension (p.001), indicating that affective computing technology effectively enhanced the emotional connection between users and AI. Additionally, the affective computing AI group achieved a course completion rate of 85%, significantly higher than the 62% in the traditional AI group (p=.01). Discussion The study results demonstrate that AI systems integrating affective computing technology can significantly enhance the quality of therapeutic relationships in mental health education scenarios, with effect sizes comparable to those achieved by human therapists in similar short-term interventions. The findings indicate that AIs ability to capture and respond to emotions is crucial for establishing "human-machine empathy." This technology not only increases users sense of trust but also promotes adherence to educational tasks. Future digital mental health intervention tools should not only focus on the professionalism of the content, but also on emotional intelligence during the interactive process. Funding No. LJ112510162011; No. 2021LNZYZD008.
Duan et al. (Sun,) studied this question.