Chatbot Efficiency—Model Testing

The factors hypothesised in the conceptual model of chatbotChatbot efficiency (see Chap. 4 , Bialkova, 2024a) were tested in an empirical study. UsersUser who had used a chatbotChatbot at least once in their life were invited to complete a survey and to provide their opinion about the experienceExperience they had with the chatbotChatbot. 90% of our respondents have employed a chatbotChatbot to contact the customer service, showing the growing importance of AIArtificial Intelligence (AI) systems in substituting human agents at the front service line. The results from the regression modelling clearly show the relationships between the factors hypothesised in our conceptual model. (1) The greater the qualityQuality and the ease of useEase of use were perceived to be, the higher the satisfactionSatisfaction was and the more positive the attitudesAttitudes toward chatbotsChatbot were. (2) The higher the satisfactionSatisfaction was, the greater was the intention to use a chatbotChatbot and the higher the willingness to recommend it. The same tendency emerged for attitudesAttitudes. (3) Enhanced functionalityFunctionality led to a more positive evaluation of chatbotChatbot qualityQuality and ease of useEase of use. (4) EnjoymentEnjoyment also emerged to play a role in perceived qualityQuality and ease of useEase of use. Note, however, some of the above parameters may turn into barriers. Although the satisfactionSatisfaction level was relatively good, consumers who are not satisfied with a chatbotChatbot will not use it in future. Such outcome is a warning call to look for appropriate techniques for assembling machine learningMachine Learning (ML), natural language processingNatural Language Processing (NLP), and reasoning to build better systems, prioritising a human-centred approach.