Modelling a conversational agent (Botocrates) for promoting critical thinking and argumentation skills

Students in higher education institutions are often advised to think critically, yet without being guided to do so. The study investigated the use of a conversational agent (Botocrates) for supporting critical thinking and academic argumentation skills. The overarching research questions were: can a conversational agent support critical thinking and academic argumentation skills? If so, how? The study was carried out in two stages: modelling and evaluating Botocrates' prototype. The prototype was a Wizard-of-Oz system where a human plays Botocrates' role by following a set of instructions and knowledge-base to guide generation of responses. Both stages were conducted at the School of Education at the University of Leeds. In the first stage, the study analysed 13 logs of online seminars in order to define the tasks and dialogue strategies needed to be performed by Botocrates. The study identified two main tasks of Botocrates: providing answers to students' enquiries and engaging students in the argumentation process. Botocrates’ dialogue strategies and contents were built to achieve these two tasks. The novel theoretical framework of the ‘challenge to explain’ process and the notion of the ‘constructive expansion of exchange structure’ were produced during this stage and incorporated into Botocrates’ prototype. The aim of the ‘challenge to explain’ process is to engage users in repeated and constant cycles of reflective thinking processes. The ‘constructive expansion of exchange structure’ is the practical application of the ‘challenge to explain’ process. In the second stage, the study used the Wizard-of-Oz (WOZ) experiments and interviews to evaluate Botocrates’ prototype. 7 students participated in the evaluation stage and each participant was immediately interviewed after chatting with Botocrates. The analysis of the data gathered from the WOZ and interviews showed encouraging results in terms of students’ engagement in the process of argumentation. As a result of the role of ‘critic’ played by Botocrates during the interactions, users actively and positively adopted the roles of explainer, clarifier, and evaluator. However, the results also showed negative experiences that occurred to users during the interaction. Improving Botocrates’ performance and training users could decrease users’ unsuccessful and negative experiences. The study identified the critical success and failure factors related to achieving the tasks of Botocrates.