Promoting Higher Order Thinking Through Reflection with a Generative AI Chatbot (Stage 2, 10:35 AM)

• In Event: e-Lightning Ed-Talk Session 2 - Stage 2

Abstract

We investigated how a generative AI chatbot could augment learning from instruction by encouraging higher order thinking (HOT) about the mathematics content through post-assessment reflection. Our AI chatbot, Math ThinkBot, was designed to prompt children to explain and compare multiple strategies they produced on ratio problems. Nineteen fifth and sixth graders interacted with Math ThinkBot after assessment. Results showed that Math ThinkBot produced HOT talk (HOTT), indeed more often than children, but also children’s use of HOTT was positively correlated with their math performance. In some cases, self-correction and conceptual insight emerged during dialogue. These findings suggest that GenAI can be a scalable, low-level scaffold for metacognitive engagement and conceptual understanding in math when designed as a reflective partner.

Authors

• Banu Karyagdi, University of California - Irvine

• Hongyang Zhao, University of California - Irvine

• Shuyuan Yu, University of California - Irvine

• Ella Rose, University of California - Irvine

• Shayan Doroudi, University of California - Irvine

• Katherine T Rhodes, University of California - Irvine

• Lindsey E. Richland, University of California - Irvine