This paper discusses preliminary observations on the creation and implementation of a GPT designed to provide constructive and reflective feedback to second-year EFL students on academic writing. It also explores how students engage with dialogic feedback provided by AI and revise their texts based on it.

The biggest issue with using AI for writing feedback is that these agents tend to completely rewrite any input they receive. This action robs students of the opportunity to learn through constructive and reflective feedback, as they would, for instance, in a peer tutorial session at a writing center.

The GPT was designed according to the theories and practices of writing center tutoring, with the goal of providing students with constructive and reflective feedback on their writing without rewriting their texts.

The participants were sixteen CEFR A2–B1 second-year university students in Japan enrolled in an English academic writing course. After writing a five-paragraph essay, students were given one week to use a GPT created by the instructor to help them revise their essays. The researcher compared the first and second drafts and analyzed transcripts of their conversations with the GPT.

The paper highlights both the potential of dialogic AI tutoring to support process-oriented writing instruction and the challenges students face in critically engaging with automated feedback.

Authors:
Diego Oliveira, Nihon University, Japan

About the Presenter(s)
Diego Oliveira is currently an assistant professor in the College of International Relations at Nihon University in Japan.

See this presentation on the full schedule – Saturday Schedule

Bookmark