Evaluating students’ perceptions, writing outcomes, and challenges in using AI-based feedback systems for english academic writing

Read full text
Authors: Sri Wahyuni, Desi Surlitasari Dewi, Rini Susanti Wulandari, Thohiriyah, Didit Kurniadi, Fani Hindun Muawanah, Nadya Zulfa Chairunnisa, Mega Aulia Ardhi
Cited by: 0
References: 21

Abstract

The increasing prevalence of artificial intelligence (AI) in higher education has transformed student engagement with academic writing, especially via automated feedback systems. This study seeks to investigate undergraduate students' experiences with AI-driven feedback tools by analyzing three primary dimensions: perceived utility, usability, and system dependability. The study employed a mixed-methods approach and included 156 English Education undergraduates from two Indonesian institutions who often utilized multiple AI-assisted platforms, including Grammarly, ChatGPT, Quillbot, and Microsoft Copilot. Quantitative data were collected using a five-point Likert-scale questionnaire, while qualitative insights were derived from open-ended responses aimed at capturing personal perspectives on benefits and obstacles. The results indicate that students predominantly possess favorable views on AI-assisted feedback, with usefulness attaining the highest rating (M = 4.24, SD = 0.67). Students indicated enhancements in grammatical precision (M = 4.32), vocabulary sophistication (M = 4.21), and increased understanding of syntactic structure and lexical selection. The ease of use received a favorable rating (M = 4.18), mostly attributed to the accessibility and immediacy of automatic response. Nonetheless, reliability received a marginally lower grade (M = 3.91), as students observed sporadic mistakes, inconsistent recommendations, and a restricted comprehension of context or rhetorical meaning. Qualitative responses underscored difficulties including excessive dependence on AI corrections, ambiguous error explanations, and inadequate help for advanced writing elements such as idea development and coherence. The study indicates that AI-driven feedback is a beneficial adjunct to writing training, providing efficiency and linguistic assistance; yet, it cannot entirely supplant human feedback in managing content arrangement and more profound conceptual requirements. An equitable integration of AI technologies with instructor facilitation is crucial to optimize their educational influence on students' academic writing advancement.

Fields of Study & Subjects

Artificial Intelligence in Education EFL/ESL Teaching and Learning English Language Learning and Teaching

Publication details

Journal / source
Multidisciplinary Reviews
Volume
9
Issue
8
Pages
2026380–2026380
Publication date
16 January 2026
Language
EN