Building an emotional IPA through empirical design with high- school students

As suggested by Woolf, one of the Artificial Intelligence (AI) grand challenges in education is "mentors for every learner". Mentors can be implemented as Intelligent Pedagogical Agents (IPAs) in order to motivate students: "The first way such systems [IPAs] must evolve is to directly address 21st century skills such as creativity, critical thinking, communication, collaboration, information literacy, and self-direction." AI provides the tools to build computational models of student's skills and scaffold learning. Further, AI methods can act as catalyst in learning environments to provide knowledge about the domain, students, and teaching strategies through integration of cognitive and emotional modelling, knowledge representation, reasoning, natural language question answering, and machine learning methods. In this paper, we describe the design and development of an Intelligent Pedagogical Agent that will guide the students through the use of a Serious Game implemented to teach STEM subjects. The use of IPAs is proposed as a support during the game evolution because they act as learning facilitators and guide the learners in the virtual environment by explaining topics, answering questions, giving feedbacks, helping the learners to collaborate with others, providing personalized learning support. The IPA that we propose in this paper behaves along two paths: Emotional, as the interaction depends on the feelings of the student, and pedagogical, i.e., by giving tips and advices related to the topics and to the tasks assigned. The emotional part of the IPA has been built conducting a study on emotions with high school students. On top of the results coming from the study, it has been built an Android application that uses the IPA as a standalone application to prove the efficacy of the IPA itself.