Learning through Generative Questioning with Generative AI: An ICAP approach

• In Event: AERA Poster Session Friday 7:45 am
  In Poster Session: ATL Poster Session 2: AI Integration and Outcomes in Postsecondary Settings

Abstract

This study investigates how undergraduate biology students engage with generative AI (ChatGPT) through self-directed questioning. Using the ICAP framework, we analyzed 784 student-chatbot interactions to examine question types, transition patterns, and differences across performance levels. Results show that longer conversations foster more Constructive and Interactive questioning, especially among higher-performing students. Passive questioning dominated shorter exchanges and lower-performing students. While ChatGPT responses were accurate and specific, their instructional helpfulness varied. These findings highlight generative questioning as a key marker of meaningful learning with large language models.

Authors

• Lu Ding, University of South Alabama

• Rachel Rodenberg, University of South Alabama

• Erkan Er, Middle East Technical University

• Ha Nguyen, University of North Carolina - Chapel Hill