Supporting Learner Agency in Collaborative Writing with Generative AI
Saved in:
| Title: | Supporting Learner Agency in Collaborative Writing with Generative AI |
|---|---|
| Language: | English |
| Authors: | Sujin Kim, Hyo-Jeong So, Kyudong Park |
| Source: | British Journal of Educational Technology. 2026 57(4):984-1008. |
| Availability: | Wiley. Available from: John Wiley & Sons, Inc. 111 River Street, Hoboken, NJ 07030. Tel: 800-835-6770; e-mail: cs-journals@wiley.com; Web site: https://www.wiley.com/en-us |
| Peer Reviewed: | Y |
| Page Count: | 25 |
| Publication Date: | 2026 |
| Document Type: | Journal Articles Reports - Research |
| Education Level: | Higher Education Postsecondary Education |
| Descriptors: | Personal Autonomy, Collaborative Writing, Artificial Intelligence, Natural Language Processing, Computer Interfaces, Writing Strategies, College Students, Writing (Composition), Semantics, Man Machine Systems |
| DOI: | 10.1111/bjet.70015 |
| ISSN: | 0007-1013 1467-8535 |
| Abstract: | To support learner agency in LLM-powered writing environments, this research introduces a new interface that integrates two key features: an "Argument Outline" and a "Similarity Viewer." These features were designed to enhance intentionality in writing and self-monitoring of reliance on AI-generated suggestions. Implemented in CoAuthor, a GPT-based writing tool, the interface was tested in an experiment with 52 university students who completed an argumentative writing task under two conditions: (1) SV group (n = 26) with Similarity Viewer only, and (2) SV-AO group (n = 26) with both Similarity Viewer and Argument Outline. Writing behaviours such as seeking, dismissing, accepting and modifying AI suggestions were analysed using statistical methods and epistemic network analysis (ENA). The average semantic similarity was calculated to measure the similarity between final texts and AI-generated content. Follow-up interviews were conducted to explore participants' perceptions of the two features. Results showed that the SV-AO group had stronger centrality in behaviours linked to independent text composition and produced final texts with lower semantic similarity. Interview data further indicated that participants viewed the Argument Outline as a useful scaffold for supporting independent writing while enabling the selective integration of AI-generated suggestions aligned with their writing goals. Overall, this study highlights the potential of interface design in optimizing the balance between learner agency and AI assistance in human--AI co-writing. |
| Abstractor: | As Provided |
| Entry Date: | 2026 |
| Accession Number: | EJ1508514 |
| Database: | ERIC |
Be the first to leave a comment!
