Sharing Secrets with Robots

摘要

This paper presents initial findings from a large-scale study that evaluated levels of student disclosure on sensitive topics. Four different conditions of survey delivery were applied and follow up interviews were undertaken. Non-parametric tests were used due to the data not satisfying the assumptions of parametric statistical tests; Wilcoxon Signed Ranks tests were conducted to examine the differences further. Preliminary data suggest that students disclosed additional information to the chatbot on more sensitive topics when the length of engagement was increased, but that this effect could be negated by the inclusion of the depth of engagement questions. Such findings suggest that the sensitivity of the student-chatbot conversation is critical in determining the influence of the chatbot, and that particular care should be taken when designing contextually-relevant ‘icebreaker’ questions.