ChatGPT in Dissertation Writing: Balancing Innovation and Rigour

Abstract

This study examines ChatGPT integration in master dissertation writing with a double impact: improving surface fluency at the expense of sacrificing critical thinking and intellectual honesty. By bringing together the voices of master’s students and the experiences of their supervisors, it sheds new light on how ChatGPT is shaping dissertation writing. Grounded on mixed methods, this study sent out 210 questionnaires to Master's students and collected extended interviews with 12 academic supervisors from three Algerian universities. Introducing the concept of the Efficiency–Engagement Paradox as a novel lens, it highlights how ChatGPT simultaneously facilitates linguistic fluency while disrupting reflective engagement in graduate-level scholarship. Results show that 72% of the students find ChatGPT helpful for planning and coming up with ideas, but 64% expressed concern about a decline in critical thinking and reflective engagement, and 48% had no idea about AI policies at universities. Supervisors observed a decline in student autonomy, weaker reasoning skills, and a shift away from interactive, dialogue-based supervision. This research contributes an empirical and theoretical perspective by focusing on dissertation writing, a rarely studied context in AI literature, and recommends open AI policy, combining AI literacy education, and process regulation to facilitate ethical, critical, and pedagogically sound use of generative AI in universities.