News
Organizational Matters: LLM Usage, Getting Useful Feedback, Invitation to Eye-Tracking Study
Written on 11.08.2025 14:41 by Norman Peitek
Dear seminar participants,
we have a few organizational announcements for you.
1. Usage of LLMs for Seminar Paper
Some of you are already busy drafting your seminar papers and reached out with questions. We would like to clarify that language correction and translation tools can be used. However, we would like to highlight that the content of the seminar paper must be your own work. If entire paragraphs are generated by LLMs, we cannot longer consider it as your own work and, like having the paper written by a third party, will result in failing the seminar.
We are not writing this news item as a preventive measure, but from our past experience, where some drafts are largely generated and then use the supervisors’ feedback only as a prompt for a revised version, but not to critically assess your work and learn from it. This is unproductive for everyone involved and does not support achieving our learning goals. Like we have experienced in the hands-on sessions and pilot studies, human reflection and learning is still quite necessary. This applies the same way to writing a seminar paper, in which you should critically reflect on your own approach and take our feedback as a way to improve your scientific writing.
Our seminars are structured in a way that you develop the skills needed in the future to assess the output of LLMs. But, at this point we cannot assume that they are already established and must insist on students' producing their own work. We encourage you to see the seminar as an objective test of your own knowledge and skills on a topic, rather than trying to make your way to the end by using an LLM to produce a questionable paper.
2. Tips for Getting Useful Feedback
As mentioned in the kick-off session, our experience has shown that students asking for feedback in a productive and early manner tend to have better grades. Please use the opportunity to learn from and with your advisor. So how do you successfully collaborate with your advisor?
- Ask specific questions. For example, if you are unsure about a decision, section, or figure, briefly present your thought process in an email or as an annotation in the PDF. Specific questions can be answered in much more detail than a general approval "is this paper okay?" question.
- Your advisor is for scientific questions, not a proof-reader or grammar check. Make sure to give your paper a high-level check before sending it to your advisor, otherwise they might be distracted and miss more fundamental issues in your paper.
- Ask for feedback early. Providing detailed feedback, and incorporating that feedback properly, takes a lot of time. This means start your work well ahead of the deadline (like many of you already did!) and plan with a time buffer at the end.
- Your advisor has a life, too. Reading 10-15 pages and providing detailed feedback takes time and energy. Many advisors are heavily involved in teaching SELab for the upcoming weeks and may already have a full day. Do not expect an immediate reply and consider that in your project timeline.
3. Invitation to Eye-Tracking Study
Finally, we would like to invite you to a study at our chair. We are conducting a short eye-tracking study on how type annotations help us understand code. You have learned how much effort it is to conduct studies, so now you can help us and sign up directly here: https://calendly.com/annotations-study/eye-tracking-study-1
Participating in studies is a good way about learning about empirical research and an excellent preparation for conducting your own research in the future. We would appreciate your help!
Best regards,
AICA-HoGC-Team