Exploring Generative AI in Education: Large Language Models for Topic Modeling and Discourse Analysis

• In Event: Virtual Poster Session
  In Virtual iPresentation Only: Division G - Section 5: Inquiry, Transformation, and Communities Virtual Poster Session

Abstract

This study explores how generative AI is discussed in the educational landscape by applying large language models (LLMs) to perform topic modeling on a large-scale corpus of social media data. We collected over 200,000 tweets related to generative AI in education and used GPT-based embedding techniques to conduct topic modeling using BERTopic. Our analysis prioritizes interpretability and data-driven discovery. The findings reveal evolving concerns around ethics, academic integrity, and instructional adaptation. We demonstrate that LLM-powered topic modeling offers scalable, interpretable insights into how educators and learners engage with generative AI.

Authors

• Wanli Xing, University of Florida

• Fan Zhang, University of Florida