Development of Semantic Differential Scales for Artificial Intelligence Agents

초록

The purpose of this study is to develop a comprehensive measurement scale for analyzing human attitudes toward AI agents by employing a semantic differential scale. This is considered to be an effective method of evaluating both functional and emotional features. Previous studies have only limitedly examined human responses to AI agents without considering objective functionality and users' subjective emotions concurrently, despite increasing human interaction with AI agents. This study applies semantic differential scales as the measurement model to measure multi-dimensional concepts of human attitudes toward AI agents. An online survey was conducted by a professional online survey agency from October 15 to 19, 2021. In the study, 55 bipolar adjectival pairs were presented to 561 respondents, from which 517 responses were gathered for final analysis after excluding unreliable responses. The results show that respondents' attitudes toward AI agents consist of six factors: intelligence, activity, expressiveness, usability, mood, and naturalness. This study contributes to research on users of AI agents by providing a comprehensive tool to investigate attitude toward AI agents and shedding light on how to identify and construct evaluation criteria for AI agents.

키워드