Examining the effectiveness of artificial intelligence tools in enhancing undergraduate students' research and academic writing skills

Abstract

The growing use of artificial intelligence (AI) tools in higher education has substantially influenced undergraduate students’ research and academic writing practices. However, existing discussions often remain broad and descriptive, frequently combining student use, teacher readiness, policy, and ethics without clearly identifying the central construct under investigation. This study therefore focuses specifically on undergraduate students’ perceived effectiveness of AI tools in research and academic writing within a Research Methodology course. Using a cross-sectional survey design, data were collected from 39 purposively selected undergraduate students enrolled in the Bachelor of Teaching Arabic as a Second Language programme at the Kulliyyah of Education, International Islamic University Malaysia, during Semester 2 of the 2025/2026 academic session. The questionnaire included demographic items, 35 Likert-scale items, and one open-ended section. The findings indicate that students perceived AI tools as useful for literature review, idea generation, paraphrasing, proofreading, readability improvement, and draft development. ChatGPT, QuillBot, Gemini, and Grammarly were the most frequently used tools. At the same time, students expressed reservations regarding trust, plagiarism prevention, and over-reliance on AI, while emphasising the need for ethical guidance, AI literacy training, and clearer institutional policies. The study contributes by offering a more focused conceptual framing of AI-assisted writing through five dimensions: research efficiency, writing quality enhancement, idea development, reference support, and academic confidence. Nevertheless, the study remains exploratory due to its small sample size, limited statistical analysis, and the absence of reported internal consistency reliability and construct validity testing. Future research should involve larger samples, stronger psychometric evaluation, and more advanced inferential analysis.