“Cognitive Co‐Pilot” or “Ghostwriter”?—The Usage Mechanism and Adjustment Strategies of Learners for Generative AI in Academic Research

ABSTRACT Generative Artificial Intelligence (GenAI) has rapidly penetrated academic research, with both advantages such as efficiency improvement and cognitive supplementation, as well as challenges such as inaccurate content and ethical risks, triggering a positioning controversy between “cognitive co‐pilot” and “ghostwriter”. Existing research lacks exploration of the usage mechanism to address this positioning controversy from the learner's perspective. This study adopted a qualitative research method based on grounded theory, conducted semi‐structured interviews with 18 learners covering educational levels from undergraduate to doctoral degrees and multiple disciplinary fields, and performed four‐level coding analysis combined with NVivo software. The results show that learners are driven by the advantages of efficiency perception, cognitive supplementation, and academic need matching, face constraints from technical limitations, low disciplinary adaptability, ethical risks, and cognitive dependence, and achieve adjustment through strategies of tool optimisation, content validation, and cognitive reconstruction. Finally, this study constructs a usage mechanism of “advantage‐driven—dilemma‐constraining—adjustment strategies”, and condenses five essential modes of existence of GenAI in academic research, namely tool‐based, differentiated, contradictory, fluid, and reflective, combined with the characteristics of cognitive and usage patterns. This study fills the limitation of existing research that emphasises technology over users, and provides empirical evidence for learners to use GenAI while adhering to subjectivity and academic ethics, for differentiated guidance by educators, and for developers to optimise disciplinary adaptability.