An AI-assisted draft cannot show who did the thinking
A first-year university study examined 962 prompts from 105 computing students. Their prompt journals offer a view of how they used AI during different parts of a task. The digest urges attention to the person’s choices, not a label based on prompt style. This is not proof of learning or a K-12 grading guide.
What the source reports
Ali Shakiba and Thomas Chaffey studied AI use by first-year university computing students. The digest reports 962 prompts from 105 students and describes the work as a prompt-journal study. A prompt can show a request, but it does not by itself show what the student understood or decided. The study points toward looking at the person’s part in each phase of work, rather than giving a student a fixed label based on prompt style. This is evidence from university students in computing, not a trial of a new classroom rule. The prompts are not proof of learning, and the study does not supply a grading guide for younger students. A polished AI-assisted answer can leave the learner’s choices hard to see.
Original source
- Title
- Who Is Doing the Thinking? Characterising Human Agency in First-Year Students' AI Interactions
- Author
- Ali Shakiba, Thomas Chaffey
- Publication
- arXiv
- Date
- Tuesday, November 3, 2026