Assessing expectations and potential of domain independent corporate learning chatbots

Abstract

Advancing technological developments in Natural Language Processing have led to a new generation of conversional agents for various application domains. A special challenge in the area of education is to handle the different domains of knowledge in order to enable domain independent learning with chatbots. To enable a future development of such a system, we investigate the different expectations of users and their satisfaction when learning with a chatbot through a Wizard-of-Oz study. Subsequently, we identified and categorized intentions of learners. The study shows high acceptance and willingness to further interact with the system. Technical requirements can be derived from our results to enable a domain-independent chatbot for corporate e-learning in the future.