PlagiarismBot or writing buddy?

“How am I supposed to use AI when writing my thesis?” This question comes up regularly in coaching sessions at the TU Delft Writing Centre where I work as a writing coach. I usually respond with a question of my own: “How do you use AI right now?” The answers differ. Some students don’t use AI at all. They prefer to do everything themselves, don’t really know how to use it, or deliberately avoid it because of the social, economic, or ecological consequences of AI use. Others jot down their ideas in no particular order and then ask AI to turn them into a coherent text. Most students fall somewhere in between. Once they’ve explained their approach, they almost always ask: “Is that allowed?”

For a long time, the rules were relatively straightforward. If you copy something word for word, you cite it, otherwise it’s plagiarism. You can paraphrase, as long as you reference the source. Having someone else write your text is fraud, getting feedback from a fellow student is not. With AI, that clarity has largely disappeared. Letting a language model write an entire text and then submitting it as your own is clearly fraud. But beyond that, things quickly become murky. What if a student is stuck on the phrasing of a sentence and asks AI for five alternatives, then uses one of them? Is that just a more advanced thesaurus, something you can use without referencing? Or is it copying text from another source, which would require a citation? And what about using AI to correct grammar or suggest a different paragraph order? Or, controversial but common, to rewrite a text in a more academic tone? A large grey area has emerged.

I’ve met many disappointed students who refused to use AI and are graded with sixes

Students get mixed messages. There is a lot of discussion about AI in education, and most supervisors allow its use. Students are also told that AI is part of the future, and that learning to work with it is essential if you ever want a job. At the same time, they are warned about the risks of plagiarism. But, at least at my own faculty, it is not always clear to students where the boundaries actually lie.

Research (in Dutch) now shows that job applications written with the help of AI are rated more positively by recruiters. In coaching sessions, I hear a similar pattern. I’ve met many disappointed students who refused to use AI and are graded with sixes, while people around them using AI get sevens and eights. Even if that difference can’t be attributed to AI alone, it is hard to stay motivated when you have spent hours struggling with a text while others polished theirs in minutes.

The problem isn’t AI, it’s the grey area around it. Students should be able to learn how to use AI as a tool without constantly worrying that they are committing fraud. At the same time, they still need to learn how to think and write for themselves. That requires clear guidelines. Not a panic-driven decision to ban AI altogether, but guardrails that allow students to experiment without fear of plagiarism and with a level playing field. As long as these guidelines are missing, every student will keep asking the same question: “Is this allowed?”