Adapting Tutoring Behavior to Student Needs A Study in Intelligent Tutoring Systems

Students greatly benefit from personalized one-onone tutoring, but resources do not usually exist to provide each student with a human tutor. For this reason, Intelligent Tutoring Systems (ITS) are an attractive way to provide students with the personalized help they need to learn without straining teacher resources. As people tend to ascribe more human characteristics to robots than purely digital characters, and are more likely to follow their instructions, robot tutors, in particular, are a promising way to provide more students with the benefits of personal tutoring. While many ITS currently provide recommendations for further study material or can generate hints to complex problems, they do not yet deliver the level of personalized help of which human tutors are capable. Human teachers can tailor their actions and help strategies to individual students, so in order to best emulate them, robots must also be able to adaptively change their behaviors to suit the needs of students. This paper proposes the design of a study to test such an adaptive robot tutoring agent and presents the implementation of an adaptive tutoring system. The platform of the tutoring system includes a robot and a connected tablet application, and will be used in the study. Students will do math problems on the tablet while getting assistance from the connected robot. The robot will, based on the decisions of the model, provide one of its possible tutoring behaviors (including giving a hint, a worked example, a interactive tutorial, or a break to the student) after each attempt to answer a question.