Analysing User Interaction with AI Chatbots in Academic Libraries

Abstract

The recent pace of change in the digitization of Information Services in Academic Libraries has been enforced by the introduction of some Artificial Intelligence (AI) technologies which rely on the use of intelligent chatbot systems for delivering automated reference information, information retrieval and virtual user assistance services. The present study was peer-reviewed, and its research problem focused on understanding how the users interact with the AI chatbots in academic libraries and discovering how problematic, emotional, or technological issues can affect levels of user satisfaction, trust, participation, and efficiency with respect to the use of AI chatbots. The research methodology used in this research was quantitative, and the data collected were structured questionnaires, chatbot verbatim log and behavioural data from the system of students, postgraduate students, research scholars, faculty members, and regular users of the academic library. The proposed model is innovative with the incorporation of new indices in order to assess the interaction effectiveness between the user and the chatbot in a multidimensional fashion, such as Dynamic Conversational Trust Index (DCTI), Conversational Efficiency Ratio (CER), Sentiment Interaction Score (SIS) and Integrated User Interaction Index (IUII). Each of the findings from the statistics showed significant positive relationships between chatbot response, perceived accuracy, trust, conversational quality, and satisfaction. User Trust (UT) had the highest positive correlation with User Satisfaction (r = 0.84, p < 0.001), and regression analysis showed that the DCTI has a significantly effect on user satisfaction ((\beta = 0.42, p < 0.001). Moreover, a high classification accuracy of 91.4%, precision of 0.89, recall of 0.92, F1-score of 0.90, and a ROC-AUC score of 0.93 were obtained from the Random Forest Classification model. Based on the findings, AI chatbot systems have a significant positive impact on academic library services, particularly in accessing library resources, speeding up response time, delivering digital reference, and fostering increased engagement in modern learning contexts.