The Impact of <scp>GenAI</scp> on Epistemic Construction, Friction and Regulation: Doctoral Researchers' Duoethnography

ABSTRACT This study focuses on the process by which doctoral researchers construct epistemic sovereignty and control in AI‐assisted writing. This study employs a duoethnographic approach to analyse the reflective dialogues of two multidisciplinary doctoral researchers. It examines their use of ChatGPT and DeepSeek for academic writing through a theoretical framework of self‐training, epistemic construction, and epistemic friction. The main findings are as follows: (1) The quasi‐subjective language output presented by GenAI tools blurs the academic identity boundaries of researchers, prompting doctoral researchers to reevaluate their initiative in writing; (2) There is a significant tension between the synchronous generation logic relied upon by GenAI and the inherent diachronic reasoning methods of human beings, revealing the epistemic contradiction between the pursuit of efficiency and in‐depth exploration; (3) Although the standardised responses of GenAI may potentially suppress academic innovation, they can also inspire researchers to develop a reflective resistance strategy, thereby promoting epistemic creation and breakthroughs. Based on the findings, this study explains the core process of how doctoral researchers, in human‐computer interaction, can activate a self‐regulation mechanism to cope with epistemic friction and, in the long run, achieve epistemic symbiosis and the production of original knowledge.