Student Evaluative Judgements of Writing and Artificial Intelligence: The Disconnect between Structural and Conceptual Knowledge

Abstract

This paper reports on how undergraduate students evaluated writing outputs created with and without generative artificial intelligence (AI). The paper focuses specifically on two aspects of writing and AI: how prior writing knowledge influenced students’ thinking about AI tools, and how the writing skills to which they were exposed in the writing classroom helped them work with AI-generated materials. This research builds upon Bearman et al.’s (2024) work on evaluative judgement as a pedagogical tool to support learners as they work with AI-mediated texts. The paper uses this lens to identify challenges that learners have in applying writing knowledge to AI-mediated situations and to devise pedagogical means to support student learning in these contexts. We found that, while students could typically evaluate structural components of writing, they struggled to evaluate conceptual ideas both for AI and human generated texts. The findings speak more generally to the need for students to develop their evaluative abilities, as well as ways that AI may reveal and amplify existing challenges that learners have with evaluating the quality of writing, engaging with source materials, and applying genre knowledge to create meaning.