This study aims to investigate doctoral students’ Generative artificial intelligence (GenAI) literacy in scientific article writing, explores their perceived roles, responsibilities, agency and ethical positioning in Human–AI Collaborative Writing (HAICW), and examines the alignment and misalignment between their demonstrated GenAI literacy and perceived HAICW practices.
An explanatory sequential mixed-methods design was employed involving 62 doctoral students from diverse academic disciplines engaged in AI-assisted scientific writing. Quantitative data were collected through a GenAI literacy assessment, followed by semi-structured interviews to deepen interpretation. Cross-case comparison was conducted to identify patterns of alignment and misalignment between literacy levels and perceived HAICW practices.
Doctoral students demonstrated high GenAI literacy (M = 7.88; Med = 8; s = 1.51; Q = 3). However, high literacy levels did not consistently correspond with a clear understanding of roles, responsibilities, agency and ethics in HAICW. Three exploratory typologies were identified, representing varying degrees of alignment between GenAI literacy and HAICW practices. Misalignment was frequently influenced by contextual pressures rather than a lack of awareness.
The study involved a limited sample of doctoral students from a specific academic context, which may constrain generalizability. In addition, reliance on self-reported perceptions in interviews may introduce subjective bias.
The study suggests that responsible HAICW requires instructional support beyond tool proficiency. Doctoral programs should strengthen students’ epistemic competencies, ethical enactment and self-regulatory strategies through scaffolded AI-assisted writing activities. Such support may enhance epistemic agency, promote responsible AI use, and reduce overreliance and ethical risks in scientific writing.
This study contributes to the emerging literature on GenAI literacy by examining its relationship with ethical, agential and epistemic dimensions of HAICW. The findings provide preliminary insights into how demonstrated GenAI literacy may be associated with responsible HAICW practices among doctoral students in scientific article writing.
