Resumen de Timing AI intervention in the writing process for low proficiency learners: relationship with motivation, self-efficacy, and performance

Meilan Li, Kizito Tekwa

• Despite AI’s contributions to academic writing, concerns about over-reliance and diminished critical thinking skills persist, particularly among learners with low English proficiency (LEP). This study addresses the gap by investigating the writing process stage where AI intervention can best enhance LEP students’ performance, motivation, and self-efficacy. Employing a mixed-methods approach and informed by the Process Writing and Scaffolding theories, the study assessed three groups of LEP learners, each experiencing AI intervention at different writing stages—Group 1 brainstormed ideas, Group 2 brainstormed ideas and structured them into paragraphs, and Group 3 brainstormed ideas, structured them into paragraphs, and wrote a draft before seeking AI support. Findings revealed that Group 3, which engaged with AI later in the writing process, exhibited the highest improvements in motivation, self-efficacy, and performance and greater editing behaviors and confidence. In contrast, reliance on AI as a replacement tool was evident in Group 1, which engaged AI early. These findings underscore educators’ need to foster LEP student agency, independence, and confidence in writing by structuring AI intervention within traditional writing frameworks. The findings have multiple implications for theory, practice, and policies that balance traditional writing instruction with AI technology to enhance writing outcomes.