December 7, 2023: AI Tools at OVGU
2:00 p.m. | Language: German and English | Conference Room of the MPI Magdeburg, Sandtorstraße
Talks and Panel Discussion with experts from OVGU
We invited you to an intensive exchange on the topic of ChatGPT and other AI tools at the university - in teaching, for theses, for learning, for writing - our aim was to contribute to the debate and to explore how we want to and can deal with these powerful tools. Please find here the presentations and a short summary.
- Dr. Barbara Witter, Management of the Graduate Academy:
Introduction - Eduard Buzila, PhD student Data-Driven Decision Support:
Use of ChatGPT for learning and exams - Dr. Mathias Magdowski, Institute for Medical Technology
Using ChatGPT in electrical engineering
- Prof. Dr. Thomas Kahle, Professor for Algebra:
ChatGPT and me - Prof. Dr. Sebastian Stober, Artifical Intelligence Lab:
Education and training on artificial intelligence
Recap
Already while preparing this event, it quickly became clear that it is hardly possible to define a fixed guideline for dealing with "large language models" (LLM) or artificial intelligence (AI) tools in the context of science - the available tools are developing too fast, the perspectives are also too uncertain, and what seems safe to us today may already be outdated tomorrow.
All in all, we are left with a sense of serenity. While a year ago, when ChatGPT was published, there were still some very restrictive views on banning AI tools in final theses, the event on 7 December 2023 at OVGU was dominated by the opposite idea: Is it even important to explicitly name the use of AI tools - won't it soon be the case that such tools will be used for virtually every thesis anyway, and won't it seem as natural to us in a few years as the use of cars, televisions or calculators?
Some important perspectives from the event still need to be mentioned:
- LLMs will not replace human creativity - they are tools that facilitate writing, translating, rephrasing. Knowledge/scientific progress will not be replaced by such tools.
- The evaluation of the written text according to correct language, spelling and grammar will be less relevant in the future - because here the new tools help to achieve a certain basic quality without much work. However, the evaluation of the content of the scientific question and its processing, the methods, the analyses and interpretation of the results will come more into focus. This may increase the fairness of marks. Oral examinations and defences will become more relevant, as will the inclusion of the process of creating an academic thesis as opposed to simply assessing the finished manuscript.
- In the future, it will become increasingly important to have the basic skills to use AI tools and to utilise these powerful aids sensibly, taking into account the opportunities and risks. Universities will therefore (have to) teach these skills, both in regular courses and perhaps also in a kind of (compulsory?) basic course for all study programmes.
- An important goal here is to avoid a "digital divide", both among students and teachers. The aim is to achieve AI literacy so that teachers and students alike have learnt how to deal critically with AI-generated texts. The reproduction of prejudices, for example, can be an effect of LLM that must be detected and avoided.
- Universities may (have to) subscribe to licences for the use of commercial AI tools, as these tools are backed by large tech companies that make money from them. At the same time, promoting the development of open source AI tools is desirable.
Interview with ChatGPT
Thomas Kahle has conducted an interview with ChatGPT in his podcast - the professor chats with the AI, a super exciting experiment that leaves you with a little bit of goosebumps:
https://eigenpod.de/eig018-chatgpt/