HEAR what AI says: simulated empathy and supportive ChatGPT interaction for emotional well-being

Engineering notes

Engineering notes will be added by the aipentium editorial team.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为大语言模型、生成式AI、ChatGPT相关技术、计算机视觉、深度学习等高价值论文补充中文说明。

Original abstract

This study explores how simulated empathy can function as an emotionally supportive interaction feature in users’ ongoing engagement with ChatGPT for emotional support. As AI-powered chatbots increasingly appear in emotionally sensitive situations, understanding how users relate to them as emotionally supportive conversational partners becomes important. Drawing on the Stimulus-Organism-Response framework, the Technology Acceptance Model, and the Theory of Planned Behaviour, this study develops and empirically examines an integrative HEAR framework. The framework examines how users’ perceptions of empathic communication, human-likeness, and ease of use influence both cognitive evaluations and affective responses, including emotional comfort and usage disposition. Data from 892 users who had used ChatGPT to express emotions were analysed using structural equation modelling. The results show that perceived empathic communication plays a central role in shaping emotional comfort and trust, while human-likeness strongly predicts how emotionally safe users feel during interaction. These emotional factors, in turn, significantly affect users’ intention to continue using ChatGPT. Notably, users may experience ChatGPT as an emotionally supportive listener, not because it experiences emotions, but because it responds in ways that make people feel understood. By positioning empathy as more than a design feature, this study highlights ChatGPT’s emerging role in digital wellbeing contexts.

Links and sources

Comments