University Students’ Writing Dilemmas in the Age of AI : Perceptions and Realities of AI Use

This study conducted in-depth interviews with nine college students during the second semester of 2025 to analyze their perceptions and concerns regarding the use of AI in writing. The results of the analysis can be summarized as follows. First, students recognized AI not as a simple search tool but as an essential tool that deeply engages the entire writing process, from topic selection to final revision. Some students chose to write their papers on their own to preserve their thinking ability, whereas most preferred to revise and submit AI-generated drafts for efficiency. Students perceived themselves as contributing approximately 60-80% of the authorship of the final product, as they were responsible for designing and revising the prompts. Most importantly, the students did not fully recognize that AI plagiarism or editing AI text to appear human-written could constitute academic misconduct. It can be seen that students are experiencing confusion due to the lack of clear ethical guidelines on the use of AI. Second, there were no substantial differences in students’ perceptions of AI across academic disciplines; rather, their views varied depending on the nature of the course or assignment and on the instructor’s stance toward AI use. There was a strong tendency to take the use of AI for granted when coding or writing laboratory reports, whereas students who viewed writing as a process of developing thinking skills either worried that reliance on AI would deteriorate their abilities or experienced ethical conflict. Third, students expressed ambivalent emotions toward AI, because of cognitive dissonance driven by efficiency considerations and ethical dilemmas. Students sometimes asserted their authorship while distrusting the accuracy of AI plagiarism detection systems, yet some reported feelings of discomfort, unease, emptiness, and guilt when completing assignments using AI capabilities. However, they found themselves in a dilemma, feeling compelled to use AI out of fear that they would fall behind in the competition for grades if they did not. Fourth, students preferred writing instruction that focuses on the effective use of AI, rather than classes preoccupied with passing AI plagiarism checks or those that exclude AI altogether. However, paradoxically, students also felt that in actual writing classes it is necessary to create AI-free environments and to adopt assessment methods such as oral exams, in-class presentations, and discussions in order to verify students’ genuine abilities. Students wish to receive support from AI in ways that allow them to experience richer and more advanced writing without “losing themselves” in the process. Therefore, the writing process needs to be broken down into stages, with specific and explicit guidelines for how AI should be used at each step. Crucially, instructors should involve students in developing these instructional strategies so that concrete and reasonable standards reflect learners’ perspectives; only then is meaningful AI literacy education likely to be realized.