Generative AI use in EFL writing: associations with originality, critical reasoning, and metacognitive engagement in a Turkish higher education context

Generative artificial intelligence (AI) writing tools are increasingly embedded in university students’ writing practices, yet little is known about how their use relates to higher-order writing skills in English as a Foreign Language (EFL) contexts. This mixed-methods study explored associations between generative AI use and the originality, critical reasoning, and metacognitive engagement of first-year students enrolled in a public university in Türkiye. Seventy-eight undergraduates completed three timed, in-class argumentative essays with optional access to ChatGPT (idea generation), Grammarly (language suggestions), and QuillBot (paraphrasing). Data included rubric-based ratings of the essays, Torrance Tests of Creative Thinking (originality subscale), Cognitive Reflection Test scores, reflective journals, and focus-group interviews. Findings revealed two distinct usage patterns. Students who frequently relied on AI tools produced linguistically fluent and well-organised texts but displayed lower originality and less developed argumentation, often reporting cognitive offloading and diminished authorial control. In contrast, students who used AI selectively, mainly for language refinement or initial brainstorming, maintained a stronger personal voice, demonstrated greater critical reasoning, and engaged more actively in planning and reflection. These results emphasise the need for explicit guidance in AI supported writing instruction: although AI can reduce lower-level linguistic strain, uncritical dependence may limit opportunities for cognitive effort and meaningful learning in academic writing.