Towards a Generalized Framework for Intelligent Teaching and Learning Systems: The Argument for a Lightweight Multiagent Architecture

The U.S. Army’s Generalized Intelligent Framework for Tutoring (GIFT) is an important step on the path toward a loosely coupled, serviceoriented system that would promote shareable modules and could underpin multiagent architectures. However, the current version of the system may be “heavier” than it needs to be and not ideal for new or veteran ITS developers. We begin our critique with a discussion of general principles of multiagent architecture and provide a simple example. We then look at the needs of ITS developers and consider features of a general-purpose framework which would encourage message-driven, multiagent designs, sharing of services, and porting of modules across systems. Next, we discuss features of the GIFT framework that we believe might encourage or discourage adoption by the growing ITS community. We end by offering three recommendations for improvement.