Users receive recommendations from Artificial Intelligence(AI) daily. AIs are utilized to recommend a product, show the best way to get to the destination, and so on; yet, these recommendations are not always adopted by the users. Among various factor...
Users receive recommendations from Artificial Intelligence(AI) daily. AIs are utilized to recommend a product, show the best way to get to the destination, and so on; yet, these recommendations are not always adopted by the users. Among various factors that affect the acceptance of AI recommendation, self-efficacy is recently being emphasized. However, the research on the self-efficacy's effecrs on the user experience(UX) of an AI recommendation system is yet to be conducted actively. Thus, this research investigated how the self-efficacy of the users affect the UX of an AI recommendation system, focusing on the usability and the role of AI-based recommendation system. Additionally, we divided the role of self-efficacy into multiple dimensions and researched its effects by manipulating the level of risk perception following the decision-making. To do so, we conducted several experiments(N=46) through a quiz interface where an AI agent recommends a single answer. Each quiz question was provided with a varying amount of reward money which acted as an element that affects the risk perception of the participants. A pre-survey was conducted before each experiment to check each participant’s self-efficacy level. All participants were given 12 questions in total and were asked to evaluate the usability and satisfaction of the AI’s recommendation. For the evaluation, we used the ASQ(1991) asking the perceived time spent, perceived difficulty, the satisfaction of the information provided within the interface. After all experiments, a post-interview was conducted to deepen the understanding of the AI recommendation system’s role for the users. The result are the following: First, when risk perception increases, the satisfaction for AI recommendation decreases. Second, the perceived usability of the AI recommendation system differs according to the level of self-efficacy. Finally, users perceive the role of AI recommendation system during a decision-making process differently according to the level of self-efficacy and risk perception. The research provides insights into the design of an AI recommendation system that considers various aspects of users' characteristics.