Writing with Generative AI: An Exploration of Strategies Use and Writing Performance in Source-based Writing

• In Event: AERA Poster Session Friday 1:45 pm
  In Poster Session: Instructional Design, Feedback, and Learning Strategies

Abstract

This study explored students’ use of generative artificial intelligence (Gen AI) during source-based academic writing. Undergraduates (n = 243) interacted with ChatGPT 3.0 and used their textbook to write an essay for their course. We coded their conversations logs, which revealed the use of 12 strategies. Then, we used hierarchical regression to identify which strategies predicted writing performance. The results showed that task framing, AI-external knowledge use, and linguistic refinement positively predicted essay scores, whereas monitoring negatively predicted essay scores. Findings highlight how Gen AI can support or hinder writing outcomes based on strategy use, offering insights into how educators can support students’ metacognitive engagement and effective writing practices in AI-integrated learning environments.

Authors

• SeongYeup Kim, Pennsylvania State University

• Matthew T. McCrudden, Pennsylvania State University