The impacts and tensions of generative AI on doctoral students’ supervisory and peer dynamics: An activity theory analysis
Authors:
Open Access
- 23 September 2025
- journal article
- Published by Australasian Society for Computers in Learning in Tertiary Education in Australasian Journal of Educational Technology
Abstract
Doctoral students are increasingly adopting generative artificial intelligence (GenAI) tools in their daily academic activities. However, it remains unclear how GenAI influences doctoral training, particularly in terms of supervisory and peer interactions within PhD programmes. This qualitative study investigated the impact of GenAI adoption on doctoral students’ interactions with supervisors and peers within their immediate academic environments. Guided by activity theory as the theoretical framework, we conceptualise doctoral training as an academic activity system mediated by GenAI tools within specific social and cultural contexts. Through in-depth interviews and thematic analysis, this study examined the experiences of 20 doctoral students who were early adopters of GenAI at an Australian university between June and August of 2023. Two key tensions emerged from the analysis: first, the tensions arising from the dual nature of GenAI tools, characterised by their affordances and inherent limitations; second, the conflict between productivity-oriented research practices and traditional academic norms. These tensions further triggered interpersonal tensions over differing attitudes or stances towards GenAI and conflicting expectations regarding supervisory responsibilities among students, supervisors and peers. The findings reflect evolving power relations, interpersonal dynamics and academic socialisation in the context of GenAI integration. This study offers theoretical and empirical insights for rethinking doctoral supervision and training in the GenAI era. Implications for practice or policy: GenAI integration in doctoral education requires redefining the roles, responsibilities and relationships between students, supervisors and peers. Doctoral supervision should transition towards a more collaborative approach, emphasising co-learning, open communication and human-AI collaboration. Doctoral programmes need to develop clear institutional policies and structured training programmes for supervisors and students to facilitate effective GenAI use and minimise related tensions.Keywords
This publication has 43 references indexed in Scilit:
- Why students use or not use generative AI: Student conceptions, concerns, and implications for engineering educationDigital Engineering, 2024
- Empowering University Educators to Support Generative AI-enabled Learning: Proposing a Competency FrameworkProcedia CIRP, 2024
- Learning during a pandemic: an Activity Theory analysis of the challenges experienced by Aotearoa/New Zealand university studentsEducational Technology Research and Development, 2023
- Reconceptualizing ChatGPT and generative AI as a student-driven innovation in higher educationProcedia CIRP, 2023
- “I’m not the only victim…” student perceptions of exploitative supervision relation in doctoral degreeHigher Education, 2021
- Doctoral Program Types and Legitimacy of Models: Different Forms for Different Purposes2018
- Online Learning and its Users2016
- Triangulation2015
- Activity theory and individual and social transformation1999
- Grounded theory research: Procedures, canons, and evaluative criteriaQualitative Sociology, 1990