Extending Intelligent Learning Environments with Teachable Agents to Enhance Learning

Gautam Biswas, Thomas Katzlberger, John Bransford, Daniel Schwartz, & The Teachable Agent Group at Vanderbilt (TAG-V)
Tenth Internation Conference on AI in Education: AI-ED in the Wired and Wireless Future, pages 389-397, May 2001

Abstract

This paper extends our previous work on simulation-based Intelligent Learning Environments and SmartTools to computer-based Teachable Agents. Teachable Agents have been inspired by our work in classrooms where students have found it very motivating to teach and help others in problem solving tasks. These interactions also helped students to appreciate feedback and the need for reasoning in their learning processes. Our Teachable Agents do not incorporate machine learning techniques, rather, they are computer-based social agents that require explicit instruction to perform well in a given task environment. We describe the design and implementation of Teachable Agents in the domain of solving complex trip planning problems, and justify the design by preliminary experimental studies that demonstrate the effectiveness of this approach.