Navigating chatgpt in academic writing: Perceptions, challenges, and strategies among university EFL students

Abstract

This study explores EFL students' perceptions of ChatGPT's effectiveness in academic writing, the challenges they encounter, and their strategies for responsible use. Using a quantitative survey design, data were collected from 81 fifth-semester English Language Education students at Universitas Jambi, Indonesia. The instrument was a closed-ended questionnaire adapted from Al-Sofi (2024), and the data were analyzed using descriptive statistics. Most students perceived ChatGPT as effective, particularly for brainstorming (91.4%), organizing their writing (79.0%), and outlining (76.5%). Interpreted through Self-Regulated Learning theory, these results suggest that ChatGPT acts as a cognitive scaffold, most clearly during planning. However, students also expressed notable concerns. Over 70% worried about the originality of their writing and over 50% about plagiarism, while 45% to 47% feared reduced creativity and critical thinking. Through the lens of ethical literacy, these concerns show that students can recognize ethical risks but still need support in ethical reasoning and action. Regarding strategies, 85.2% believed that ChatGPT should complement rather than replace lecturer feedback, and 91.3% called for formal training on responsible AI use. Yet only 30.9% had received such training. The study concludes that ChatGPT can enhance academic writing when used responsibly. Institutions, however, must establish clear ethical guidelines, provide targeted training, and develop assessment methods that preserve critical thinking and academic integrity.