Paper Type

Complete

Paper Number

PACIS2025-1919

Description

The development of generative AI has enabled large language models (LLMs) to function as AI companions in digital communication. This study explores how differing companionship styles—specifically, the Rational versus the Emotional companionship style—influence user experience in an online scenario-based experiment. The findings show that emotional companionship style significantly enhances user satisfaction by increasing perceptions of social presence and social interactivity. However, this positive effect is reduced when LLMs explicitly display logical reasoning processes typical of a rational style. These results suggest that while emotional companionship promotes deeper user engagement, overt displays of rationality may undermine its interpersonal benefits. The study highlights the importance of balancing emotional expressiveness with cognitive clarity in the design of AI companions to optimize user experience. This research contributes new insights to the understanding of human–AI relational dynamics.

Comments

HCI

Share

COinS
 
Jul 6th, 12:00 AM

The Coldness of Your Explanation: How Emotional Intensity in Large Language Models Shapes User Satisfaction

The development of generative AI has enabled large language models (LLMs) to function as AI companions in digital communication. This study explores how differing companionship styles—specifically, the Rational versus the Emotional companionship style—influence user experience in an online scenario-based experiment. The findings show that emotional companionship style significantly enhances user satisfaction by increasing perceptions of social presence and social interactivity. However, this positive effect is reduced when LLMs explicitly display logical reasoning processes typical of a rational style. These results suggest that while emotional companionship promotes deeper user engagement, overt displays of rationality may undermine its interpersonal benefits. The study highlights the importance of balancing emotional expressiveness with cognitive clarity in the design of AI companions to optimize user experience. This research contributes new insights to the understanding of human–AI relational dynamics.