Key points are not available for this paper at this time.
Purpose This paper aims to detail how a university library developed an AI chatbot to meet a growing need for virtual reference services. This chatbot was developed using Google's free Dialogflow bot platform and embedded in the library's website. With the onset of COVID-19 and a greater reliance on virtual services, chatbots have become of increasing interest to libraries as a tool to provide enhanced services during non-staffed hours and to perform basic information triage when virtual chat transactions reach an overwhelming number of available staff. Design/methodology/approach Using in-depth research into current practices and readily available tools, a small non-technical team at a university library designed and piloted an AI chatbot that employs natural language processing and AI training. This article describes the chatbot development and implementation process. Results of chatbot interactions after one academic year of usage are also reviewed. Findings This study reveals that a university library chatbot may be developed and deployed with minimal coding knowledge using existing tools. Chatbot content can be populated through current library information sources and trained to address typical information inquiries. However, additional development and testing is needed to increase user engagement. Originality/value This study indicates that libraries can develop and deploy chatbots to meet user information inquiries without onerous technical training or IT resources. It describes best practices for chatbots and the steps necessary to deploy a chatbot on a library website.
Rodriguez et al. (Thu,) studied this question.