Designing an AI Partner for Jigsaw Classrooms

Jie Cao, Rachel Dickler, Marie Grace, Alessandro Roncone, Leanne Hirshfield, Marilyn Walker, Martha Palmer

Abstract:

The Jigsaw is a cooperative learning model that is widely used to organize
small group classroom activities so that students have to be dependent on each other
to succeed. There are a number of challenges, however, in supporting individual
students and groups of students during jigsaw activities. In this paper, we present
an AI partner designed to support jigsaw activities in middle school classrooms
through dialogue. First, we present examples of a jigsaw activity taught in our partner
classrooms, which focuses on block programming with sensors. We then share the
design of a state space and example dialogue flow for a dual-function AI partner
(supporting both group-level conversations and individual student answers) to guide
young students during jigsaw activities. We look forward to discussing how the childagent interactions that result from this design might improve both group dynamics
and student knowledge gained during cooperative learning activities such as Jigsaws.