ChatGPT in the lecture hall: How universities react to generative AI tools
Generative AI applications are not entirely new, but since the release of ChatGPT in late November 2022, no one can deny how big their impact on everyday life at universities is. Doris Weßels, Professor of Business Informatics at the Kiel University of Applied Sciences, emphasizes the great potential that generative AI has for student training. In her view, technology will transform writing processes in teaching and learning. She recommends to university teachers: enlighten, try, accept, become active. In 2022 she founded the virtual competence center “Teaching and learning to write with AI”. The online founding conference on March 23, 2023 marked the official launch of the new center.
Guidelines for faculty and students
A white paper entitled “Unlocking the Power of Generative AI Models and Systems like GPT-4 and ChatGPT for Higher Education – A Guide for Students and Lecturers” was published at the University of Hohenheim in Stuttgart in mid-March. The authors understand it as a guide for both lecturers and students.
The advice for students shows a variety of ways to improve your own learning and written work with the help of ChatGPT. The authors recommend the bot as a writing and learning partner, as a helper when revising your own texts, for summarizing extensive learning material and as support for programming. For each suggestion, the guide names concrete first steps and lists suggestions for prompts in table form to start the dialogue with the system.
The recommendations for university teachers are divided into advice for teaching and advice for examinations. Because the enormous technical possibilities of generative AI affect this part of the course particularly strongly, the authors recommend developing new examination formats quickly. Lecturers should, for example, plan an additional oral presentation for homework, encourage and ask their students to reflect on their own learning and work process, and provide for examinations in the form of websites, videos or animations.
The team of twelve authors of the white paper combines expertise in business informatics, computer science, management and sociology. Team leader Professor Henner Gimpel, who is responsible for the Department of Digital Management at the University of Hohenheim, sees opportunities for universities in the new AI tools. However, he also emphasizes that binding rules are necessary: ”One prerequisite for this are clear guidelines at every university. If the students and lecturers then use the new tools in a well thought-out, efficient and responsible manner, the entire tertiary education system could benefit to a large extent benefit. The potential for this is at least given,” says Gimpel.
In their publication, the authors also address the question of how AI tools will be used at the university if they are no longer available free of charge, as is currently the case with ChatGPT. Then it is important to avoid a situation in which some students can afford access while others cannot. In this case, universities would have to provide access so that the gap between poor and rich students does not widen further.
Housework at the push of a button?
In their outlook, the authors also formulate a warning: “If we do not consciously and purposefully integrate ChatGPT and similar tools, there is a risk that students will only use the new possibilities for passive information intake instead of developing into critical minds”.
Skeptics even fear that students could leave homework to the chatbot in the future. The experts make it clear: ChatGPT is currently not able to write complete scientific papers after entering a prompt. In addition, there is a risk with all editions that they are wrong or plagiarize other works. Students must therefore check the expenditure carefully before adopting parts of it in their work. In addition, the issues often lack details that the students have to work out themselves – in dialogue with the chatbot or in group work with fellow students. Universities will have to teach these new ways of working in the future. Students should know that they should not be blinded by the beautiful formulations of the language AI and must critically question their outputs.
The project team of KI:edu.nrw at the Ruhr University Bochum led by Peter Salden presented an evaluation of legal questions relating to AI-supported writing at the university. Professor Thomas Hoeren from the Westfälische Wilhelms-Universität Münster, who, among other things, heads the legal information center of the Digitale Hochschule NRW, discusses who is the legal author of an AI-generated text and whether there is an obligation to identify it when it is used in examination-related work.
The report clearly denies that a generative AI can be the author of a text, because ChatGPT is only a machine and not an author. Rather, students are fully responsible for their work – including any mistakes they may have made. According to Hoeren, the prerequisite for claiming authorship of a text written with the help of AI is a sufficient degree of intellectual personal contribution.
Whether there is an obligation to label depends on the applicable legal regulations. In the past, students had to submit a declaration of originality for written proof of performance, in which they explained whether and which aids they used. According to popular belief, this will also include tools such as ChatGPT in the future. In view of the uncertainty in the universities as to how to deal with the new tools, appropriate clarifications in the regulations are recommended, advises expert Hoeren.
“With the report, the legal framework for the use of AI writing tools has become much clearer,” says Peter Salden, project manager of KI:edu.nrw and head of the Center for Science Didactics at the Ruhr University. “However, it is also evident that the legal clarification is not sufficient to solve the acute questions at the universities. We must also think about what our students should learn in relation to the tools and how we can also incorporate the new possibilities into our examination formats integrate.” The KI:edu.nrw project intends to continue working intensively on these questions in the coming months.
cause for self-reflection
Some scientists see generative AI as an enormous opportunity for universities, others emphasize the challenges that systems like ChatGPT pose to the scientific community. Both sides agree that, given the enormous potential, universities need to ask themselves fundamental questions. It is about the question of the importance of writing in science and teaching and what skills universities want to convey to students in order to make them fit for a world in which generative AI will play an increasingly important role.
If the WLAN in the study is too poor, then mesh WLAN can help. In c’t 9/2023 we explain what hardware is required and how to wire the base stations. The KI Stable Diffusion can also calculate impressive images on your own computer. c’t shows how to turn your graphics card into a brush. Are you probably also the IT support in your family and circle of friends? We’ll introduce three open source remote maintenance tools and walk you through installing them on your own server. We also travel back to 2002, when Intel presented the first consumer CPU with hyperthreading, the 3 GHz “Pentium 4”. You can read that and more in c’t 9/2023!