Assessing the Potential of AI-based Tools for Scientific Writing

Schliessen Icon

Because of the broad availability and rapid development of AI-based tools such as Microsoft Copilot, ChatGPT, and Google Gemini, lecturers and students have many questions, uncertainties, and challenges: Are such tools allowed at ETH? If so, what can they be used for, and what are their limitations? How can they be used, and how can this be taught and learned?
In our project, together with the Zurich-Basel Plant Science Center (PSC) and the ETH Library, we have actively supported our lecturers and students at the Department of Environmental Systems Science (D-USYS) in using AI-based tools in their teaching and learning. The project focuses on the scientific writing process and will conclude in June 2024.

At the beginning of our project, we conducted a needs analysis among lecturers and students in a series of surveys, workshops, and individual discussions. The two most immediate needs we identified were for (1) a legal and ethical framework and (2) techniques for didactic adaptation. To address these needs early on, we first collected practical examples, curated teaching and learning materials, and created guidelines. Subsequently, we have developed a wide range of support for a total of 18 courses (BSc, MSc, PhD; course sizes of 10-150 students) at D-USYS in HS23 and FS24. These have included presentations, consultations, co-creation of course materials with lecturers, student surveys, networking/exchange events, and a Moodle course for lecturers at D-USYS, which serves as a knowledge base and networking hub.

With the right prompt and for some tasks, AI-based tools are great. Just communicate expectations well (what’s allowed, what’s forbidden, what’s expected).
Marcel Müller

What are the main components of the project?

  • To provide a framework for scientific integrity and the ethical use of generative AI that is available to lecturers and students.Based on the legal framework of ETH Zurich, we have developed scientific integrity guidelines about the responsible use of generative AI and recommendations for declaring the use of AI-based tools (e.g., textual or tabular form). We have also developed case studies and other teaching materials for moral reasoning that are available to our lecturers for easy integration into their teaching, especially into research integrity classes. They cover topics such as transparency, inaccurate and false information, attribution and privacy, confidentiality, accountability and authorship, avoiding bias. We will publish recommendations for explaining how to use AI-based tools and recommendations for unbiased prompting. Furthermore, our project partner, the ETH Library, has created a Moodle course unit about the responsible use of AI, and they also published new guidelines on “AI and plagiarism”.
  • To enable lecturers to learn about and share experiences with generative AI tools in scientific writing and to develop best practices for responsible use.This is done in the Monthly MondAI workshop series, where we provide updates about recent developments in the field, provide hands-on learning opportunities, and facilitate knowledge transfer and professional exchange. We have observed that the topic of AI is being addressed in more and more courses throughout the department.
  • To co-create custom-made course materials for selected (“pioneer”) courses. Since our project focuses on scientific writing, we recruited lecturers whose courses have a scientific writing component. We have then collaboratively designed new learning objectives, course materials, exercises, and assessment formats.
  • To address students’ concerns about AI-based tools and support their learning. In HS23, we organized the informal AI Club series, where students could discuss AI-related questions with the project team, fellow students, and representatives of the ETH Library.
  • To make our insights accessible to the ETH community. To enable other departments to learn from our insights, we have created a Moodle course for lecturers where we share best practices, recommendations, descriptions of and know-how about tools, and curated materials. Furthermore, we offered a workshop at a “Netzwerk Schreiben” event organized by the ETH Library for educational developers and scientific writing instructors from various departments and participated in the ETH Library’s Rock Your Master’s event.
It’s absolutely fine that students use AI, as long as they transparently document it.
Marlene Mader, Lecturer
Why did you launch this project?
Melanie: When we launched the project, we started with a nobody-left-behind perspective. For me as a lecturer, the challenges have been obvious from the beginning, and I wanted to ensure that lecturers can gain an understanding of these new technological developments and apply these tools in their own teaching contexts. Additionally, as an expert on research integrity, I was interested in understanding the challenges of generative AI for the responsible conduct of research, and I am looking forward to sharing our growing number of cases for moral reasoning with other lecturers in research integrity classes.
Manuel: In January 2023, I was also looking for answers to the many questions I had about the opportunities, challenges, and uncertainties of AI in teaching. However, I quickly realized that almost no one can provide me with these answers; there was simply a lack of experience, curiosity and, in many cases, courage. At the same time, it was obvious how far-reaching and disruptive this technology would be for our everyday teaching and learning (and science in general). Graduates of ETH Zurich can be expected to have all the necessary future skills for the job market and research: the T stands for technical! With the project, I could explore and find the answers I was seeking and, more importantly, share these experiences with our lecturers and students at D-USYS.
Réka: I joined the project in July 2023, just in time for the kick-off meeting. My main motivation in joining was to collaborate with others on designing learning sequences and materials that included generative AI tools. In my own teaching on scientific writing, I had already experimented with some new approaches, but the project enabled me to collaboratively create learning opportunities in a variety of learning and teaching contexts. I have found that materials tailored to the students’ most immediate learning needs help increase their engagement with this emerging topic.
What has been the impact of the project?
Melanie: The project is creating teaching materials that will be used in many classes and by many lecturers: materials for research integrity classes for the responsible use of AI and exercises that can be used by lecturers in their scientific writing classes. This tool-oriented approach is a great service to the community of lecturers at D-USYS and at ETH Zurich.
Manuel: First and foremost, rich and extensive experiences for all stakeholders. At the department, we were able to create a great deal of certainty and clarity for both lecturers and students and were able to find creative, sensible, and pragmatic ways in which we can and want to deal with AI in teaching. We have therefore created a very open, collaborative, and transparent approach to AI in teaching and can benefit from the project's many valuable and unique experiences, materials, and support services. We were also able to alleviate fears and concerns and contribute to many great developments across ETH Zurich, such as the revised declaration of originality, the networking among key stakeholders, and the interdepartmental discourse and exchange of experiences on the topic.
Réka: This project has brought generative AI closer to lecturers and students at D-USYS, and the change is palpable. We have raised awareness, improved competences, enabled exchanges, and now we can see the ripple effect occurring. Just recently we have heard about an informal student group being formed to host regular discussions on AI, initiated by a member of our Monthly MondAI series. This fills me with so much joy!
The main take-away? Learning by doing – getting to know it and knowing how/when you can use it is key.”
(anonymous student)
My key lesson: To use AI and take benefits from it but not to fully trust in them!
(anonymous student)