A Manual: Developing Artificial Social Intelligence (ASI) Lite-Scale for Service Robots

Abstract

The manifestation of “Artificial Social Intelligence (ASI)” stands as a cornerstone of a social robot system development that influences users' interactions and experiences overall in the ever-evolving landscape of user-centric HumanRobot Interaction (HRI). Recognizing the pivotal need to evaluate a socially interactive system accurately, this paper presents a unidimensional-scale measurement of ASI that measures a focused dimension of users' perceived social intelligence in a robot, minimizes participants' fatigue to generate higher response rates, maximizes the ability to conduct user-friendly research, and enhances the ease of interpreting the results that makes it more accessible to a diverse audience. Employing a cross-disciplinary literature review, personal interviews (n = 14), and large-scale surveys (n = 2,358) consisting of its five video-based stimuli data collection process, this study adhered meticulously to numerous scale measurement procedures to develop an “ASI Lite-Scale” and validated it with multiple tests, including Exploratory Factor Analysis (EFA), Confirmatory Factor Analysis (CFA), and Exploratory Graph Analysis (EGA), assessment tests of convergent, discriminant, and nomological validity, and multi-group measurement invariance analysis to establish its robustness and ability to be generalized. This study of ASI Lite-Scale provides a structured scale development manual to help fellow researchers employ this methodology and reach a wider readership, thereby fostering the development of validated scale measurements in the field of HRI