How Dynamic vs. Static Presentation Shapes User Perception and Emotional Connection to Text-Based AI

DOI

This study investigates the influence of dynamic versus static presentation modes in text-based conversational AI on user perceptions and emotional connection. We conducted a controlled, within-subjects experiment (N=103) where non-technical users interacted with a LLM-powered chatbot in both dynamic (typing-simulation display) and static (non-incremental text display) modes. Results from ANOVA showed that dynamic interactions improved perceptions of AI's competence, warmth, trustworthiness, engagement, adaptive behavior, supportiveness, personal connection, empathy, bias awareness, accountability, and emotional expressiveness. However, no significant differences were found in perceived effectiveness, bias, and learning support. These findings indicate that dynamic AI presentations can significantly enhance user experience by fostering deeper emotional connections and greater trust in the system, making AI interactions more human-like and increasing adoption of AI technologies. The paper discusses implications for human-AI interface design, emphasizing the need to mitigate risks of misinformation and manipulation of users through transparency, ethical considerations, and robust user education strategies.