נושא הפרוייקט
מספר פרוייקט
מחלקה
שמות סטודנטים
אימייל
שמות מנחים
האם אינדיקטורים של התנהגות משתמש יכולים לשקף התנהגות במערכת שופט-יועץ (JAS)
Can user behavior indicators reflect behavior in a judge-advisor-system (JAS)
תקציר בעיברית
תקציר באנגלית
The presence of recommendations on web and online platforms has become increasingly prevalent, influencing user decision-making processes. This study examines the impact of recommendation source (AI vs. human) and recommendation source credibility on user behavior and recommendation acceptance in online platforms. The study investigates how user behavior indicators, such as reaction time and clickstream, can reflect behavior in a judge-advisor system (JAS). Drawing on the decision-making literature, this research formulates hypotheses about the relationships among recommendation source, recommendation source credibility, user behavior indicators (reaction time and mouse clicks), and recommendation acceptance. The hypotheses posit that high perceived credibility of the recommendation source and AI as the recommendation source will lead to decreased reaction time and mouse clicks, indicating more efficient decision-making. Furthermore, the hypotheses propose that high perceived credibility and AI as the recommendation source will increase recommendation acceptance. In addition to the main hypotheses, the study examines in an explanatory manner the mediating role of user behavior indicators in the relationship between recommendation source/recommendation source credibility and recommendation acceptance. To empirically investigate these relationships, a 2x2 factorial design experiment was conducted with 500 participants recruited through an online platform. The participants were presented with a stock market prediction task and received a recommendation from either an AI or a human advisor, with manipulated levels of credibility. User behavior was measured through clickstream and reaction time, while recommendation acceptance was assessed using the weight-of-advice (WOA) formula. The findings of this study have implications for understanding cognitive decision-making processes and human-computer interaction in online platforms, as well as informing AI system designers.