Exploring Learners’ Self-Regulation in Collaborative Writing With AI as a Partner (Poster 10)

• In Event: AERA Poster Session Thursday 1:45 p.m.
  In Poster Session: Advanced Technologies for Learning Poster Session 1

Abstract

We examine how graduate students differ in their deployment of self-regulated learning (SRL) during collaborative writing tasks with humans versus AI (using OpenAI’s GPT-4 as the base model). We find that participants mostly employ cognitive reactions when collaborating with human partners. In comparison, they use more diverse and reflective SRL processes, such as task analysis, metacognitive strategies, self-judgment, and self-reaction, when working with AI. Analyses of participants’ interviews suggest that AI can enhance writing creativity and coherence. However, compared to human partners, the tool lacks intimacy and does not support authentic idea co-construction. Our findings illustrate the potential differences in SRL and collaboration processes when working with AI versus humans, with implications for the design of hybrid human-AI learning technologies.

Authors

• Saerok Park, University of North Carolina - Chapel Hill

• Ha Nguyen, University of North Carolina - Chapel Hill