The study, led by associate professor of computer science Sang Won Lee, employed a custom digital platform to log every interaction students had with generative AI, including prompts, responses and copy‑paste actions.

Out of the 77 participants, 71 used ChatGPT in some capacity, ranging from idea generation to text revision. Nearly half of all AI interactions involved requests for the tool to draft the entire essay, while others guided the model with step‑by‑step instructions or used it to polish their own drafts.

Researchers observed a clear split in perceived authorship: students who used AI mainly for planning or editing felt as responsible for the final work as those who wrote unaided, whereas those who let the system produce full essays reported a weaker sense of ownership.

A follow‑up experiment introduced SWAG, a monitoring platform that recorded AI use and allowed instructors to view the logs. When students knew their AI activity could be reviewed, requests for complete essays dropped sharply, suggesting accountability encourages more responsible use.

Lee’s team argues that rather than banning AI, educators should set clear limits, promote transparency and teach critical, responsible usage to preserve essential writing skills.