Generative AI Use, Self-Regulated Learning, and Critical Thinking in EFL Writing: A Correlational Survey of Indonesian University Students

Siti Nur Diana (Universitas Nurul Jadid)
Syaiful Islam (Universitas NUrul Jadid)

Jurnal Pendidikan Terapan
Vol 4, No 4 September (2026)

Purpose: This study aims to examine Indonesian university students’ patterns of generative artificial intelligence (GenAI) use, critical thinking practices, self-regulated learning (SRL) behaviours, perceived benefits and challenges, and the relationships between GenAI use, SRL, and critical thinking in EFL writing. Methods: This study employed a quantitative descriptive-correlational survey design with a cross-sectional approach. Data were collected through an online questionnaire completed by 208 students enrolled in EFL-related programs at several Indonesian universities. The data were analysed using frequency distributions, percentages, means, standard deviations, the Shapiro-Wilk normality test, and Spearman’s correlation to describe patterns of GenAI use and examine relationships among the variables. Findings: Nearly all respondents 99.0% of respondents had used GenAI for EFL writing assignments. ChatGPT was the most frequently used tool (75.0%), and planning was the writing stage in which GenAI was most commonly used (46.6%). Students reported high levels of critical thinking practices (M = 4.30, SD = 0.34) and SRL behaviours (M = 4.22, SD = 0.34). They also reported a high perceived risk of over-reliance on GenAI (M = 3.90, SD = 0.73). The frequency of GenAI use was not significantly associated with critical thinking practices (ρ = 0.033; p = 0.638). By contrast, SRL behaviours in GenAI use were positively and significantly associated with critical thinking practices (ρ = 0.449; p < 0.001). Research Implications: These findings indicate that EFL writing instruction should not focus solely on whether to allow or restrict the use of GenAI. Lecturers need to guide students in planning their use of AI, formulating purposeful prompts, verifying AI-generated information, revising outputs through their own reasoning, and maintaining academic integrity. Process-based writing instruction and reflective assessment are therefore needed to support responsible, meaningful, and self-regulated use of GenAI. Originality: This study integrates patterns of GenAI use, critical thinking, SRL, and perceived benefits and hazards into a single analytical model in the context of Indonesian EFL higher education. Also, the results demonstrate that the quality of self-regulated use is more important than the intensity of use with respect to critical thinking.