Guiding Online Students' Cognitive Presence in Asynchronous Online Discussion: Application of GPT-4.0

• In Event: Computational Thinking and Cognitive Skill Development

Abstract

This study examined how the GPT-4.0 Large Language Model (LLM) can support students in asynchronous online discussions. This study was an attempt to examine the Large Language Model Content Analysis (LACA) application through one-shot and few-shot prompts to automate the indicators of cognitive presence (CP). We compared human and GPT-4.0 coding that resulted in a moderate strength of Cohen k of .79. Results revealed that the LACA with the few-shot prompt was more accurate in automatically detecting the phases of CP in asynchronous discissions with a fair Cohen k of .33. The results of this study can guide future research to help online students become more productive and engaged in online discussions while self-detecting their own CP.

Authors

• Larisa A. Olesova, University of Florida

• Christine Wusylko, Kennesaw State University

• Bojan Lazarevic, University of Florida

• Swapna Kumar, University of Florida