Designing Emotion-Adaptive Human–AI Interfaces: An Empirical Study on Empathy, Trust, and Context-Aware Interaction

Abstract

As artificial intelligence systems become increasingly embedded in

everyday human–computer interaction contexts, user expectations regarding their

social and emotional capabilities are gradually shifting from a primary focus on

functional efficiency toward more complex dimensions such as empathetic

experience, trust formation, and contextual sensitivity. Nevertheless, despite the

notable progress achieved by large language models in terms of linguistic fluency,

such systems often remain confined to surface-level simulations of empathy. In

response to this limitation, the present study investigates an emotion-adaptive

human–AI interface that integrates real-time affect recognition, dynamic user

profiling, and context-aware response modulation within a unified framework.

Through a comparative user study encompassing task-oriented, social, and

emotionally supportive scenarios, the results suggest that emotion-adaptive

mechanisms may, to some extent, enhance users’ perceived empathy, trust, and

engagement.