Unpacking the Impacts of GAI Use on Doctoral Students’ Metacognitive Experiences in Summary Writing: A Mixed Methods Study

Generative artificial intelligence (GAI) is increasingly integrated into L2 writing, whereas its impacts on metacognitive experiences remain unclear. The mixed-methods study addresses this gap by comparing metacognitive experiences across doctoral students who chose to use GAI in a summary writing task and those who did not. Data were collected through questionnaires and stimulated recall interviews among 103 doctoral students in China. Quantitative analysis revealed no significant group differences in overall metacognitive experiences, but GAI users reported significantly lower levels of online task-specific metacognitive strategies and demonstrated greater within-group variations than non-users. Thematic analysis of interviews revealed that GAI functioned as an optional cognitive scaffold and affective regulation resource in task execution and redistributed metacognitive regulatory responsibilities across human and GAI tools, while learners engaged in ongoing metacognitive judgments of GAI use throughout the writing process. The findings illuminate the dimension-contingent and heterogeneous influences of GAI use on metacognitive experiences and offer pedagogical implications for responsible GAI use in writing.