The Development and Evaluation of a Generative Artificial Intelligence Agent: MathPal (Poster 44)

• In Event: AERA Poster Session Sunday 1:30 p.m.
  In Poster Session: Artificial Intelligence in Education and Learning Technologies

Abstract

Researchers developed a conversational AI named MathPal, following the framework of metacognitive knowledge. MathPal is designed to provide personalized support for 9th-grade algebra students as they solve math problems. This proposal aims to: 1) demonstrate the design and development of MathPal, including its conceptual and metacognitive support features; and 2) present feedback gathered from usability testing conducted by teachers and administrators. The feedback indicated that they found MathPal effective in providing conceptual and metacognitive support. They also suggested three additional features be incorporated. The development of MathPal, along with the insights gained from usability testing, has yielded design and instructional implications to inform the future design of AI chatbots in high school math contexts.

Authors

• Zilong Pan, Lehigh University

• Chenglu Li, University of Utah

• Zilu Jiang, Johns Hopkins University