VietTESOL International Convention 2026

The rapid growth of artificial intelligence writing tools has created new opportunities for university students, but it has also raised concerns about learner autonomy in English as a foreign language writing. This study examines how dependency on these tools relates to autonomous writing behaviour among engineering students at a Vietnamese university. Drawing on frameworks of learner autonomy and writing self-efficacy, it adopted a convergent parallel mixed-methods design. Data were collected through a questionnaire completed by 120 mechanical, electrical, and automotive engineering students and semi-structured interviews with 9 purposively selected participants. The quantitative findings showed that more than three quarters of the students used artificial intelligence writing tools at least three times a week, mainly to generate sentences and paragraphs rather than to revise their own writing. Tool dependency was the strongest negative predictor of autonomous writing behaviour, whereas writing self-efficacy was the strongest positive predictor. The interview data further explained this relationship through three themes: cognitive offloading, motivational displacement, and reduced metacognitive engagement. These findings suggest that unguided use of artificial intelligence writing tools may weaken students' planning, monitoring, and evaluation of their writing. The study offers evidence from a Vietnamese English for Specific Purposes context and discusses implications for curriculum design, writing assessment, and responsible technology-supported writing instruction in technical higher education.

Keywords: artificial intelligence writing tools; learner autonomy; autonomous writing behaviour; writing self-efficacy; English for Specific Purposes; Vietnamese higher education