Generative AI in Ghent University Education: Impact and Approach
Latest update: 11 September 2024
Generative AI tools impact Ghent University education, the intended competencies/learning outcomes we want to foster, and the way we teach and assess. Read the following Education Tip to learn about generative AI and the Ghent University guidelines. Take a look at the outline of options for incorporating tools into your teaching practice and learn how to promote AI literacy.
I. Generative AI: Use, Opportunities, Risks
Generative AI is a type of AI technology that can create new, original content (e.g. texts, images, sounds) based on patterns and structures it has learnt from existing data. Based on advanced algorithms and neural networks, GenAI generates output (text, images, audio, videos and computer code) that closely resembles human output.
GenAI models are trained on huge quantities of data sourced from the internet. They can generate new content statistically similar to the training data but unique and adapted to specific input or instructions. This technology has permeated various fields, including the arts, entertainment, communication, and academic research.
Want to find out more about these systems? Please take a look at the infosite GenAI: on Learning, Creating and Lecturing, Module One.
Have you never used GenAI tools such as ChatGPT before? These experimentation guidelines will lead first-time users on a step-by-step journey through this Brave New World of GenAI.
What GenAI Tools Are There?
ChatGPT (Generative Pre-trained Transformer) is the most well-known generative AI tool to date. The system was launched on 30 November 2022 by the American research lab OpenAI. It is an open access artificial intelligence writer (AI writer) that answers questions, the so-called prompts. Within seconds, the system generates an answer to your question in the form of a (brief) text. If you are dissatisfied with the result, you simply ask the system for a new answer, which will be different from the first. In addition, the system can also summarise, translate, restructure and correct texts in different languages. You can ask it to copy the references you used in your text or adopt any other reference style you prefer. Finally, the bot can also generate computer code with correct syntax in several programming languages such as JavaScript, Python and React.
There are many other chatbots besides ChatGPT, e.g. Claude, Pi, ... . Certain existing search engines now have built-in chatbots, e.g. Google’s Gemini and Microsoft’s Co-Pilot. Microsoft has integrated Co-Pilot in their other applications, like Word and PowerPoint, although only in specific licenses. In the context of academic education and research, the following AI writers are currently the most interesting ones: Consensus, Elicit, Scite_, Research Rabbit and ChatPDF. Be sure to give them all a try to find out what they are capable of.
It is beyond the scope of this Education Tip to sum up every existing GenAI tool. The website thereisanaiforthat.com tries to keep an overview and adds new apps daily. Slightly more user-friendly is the website Generative AI Tools Types and Models, where you will always find the most prevalent tools at the top of the page.
GenAI in the Teaching Practice
The possibilities for AI tools in the teaching practice are plentiful. Below you will find an overview containing a number of specific examples. Are you looking for tool suggestions per role? Download the pdf file.
Several online references offer extensive overviews of specific practical applications of AI in teaching practice in the form of case studies. Take, for instance,
- the compilation 101 creative ideas for your teaching practice,
- webpage AI voor docenten (in Dutch),
- the webpage opportunities for lecturers according to OpenAI,
- video clips containing practical examples by Harvard lecturers,
- …
Using GenAI: Risks
Using Generative AI is not without risks. Please consider the following limitations and ethical implications when using the tools:
- The use of GenAI may entail an invasion of your privacy. The free versions of the tools often process your data to train the systems further. Never feed the system sensitive data. Feeding any such system data governed by GDPR is punishable by law unless the systems have explicit rules and regulations on data processing. Please read up on GDPR and AI on onderzoektips.ugent.be.
- The tools can generate disinformation. Content-wise, the information is unreliable because it is generated based on very limited data. If the tool has insufficient data to answer a specific question (or even none at all), it will still yield a credible result. Credible but not necessarily true. In AI-speak this is called “hallucination”. It makes it difficult to judge which elements are true and which are not. These generated texts, including mistakes, fake images, etc. can start to lead a life of their own, which, in turn, contributes to the circulation of fake news.
- Only sourcing information from the samples they have been trained with, moreover, the tool’s answer/solution to a particular assignment might not at all be representative. This bias is due to a lack of data quantity and a lack of data quality. Remember that these data were originally also sourced off the internet and are thus not free from inaccuracies, prejudices, and bias.
- Normally, the developers build safety mechanisms to prevent the system from producing ethically reprehensible answers. However, these safety mechanisms can be bypassed easily by giving certain commands. What is more, ethical choices are always also culturally and ideologically coloured. At this point, with American companies dominating the field, that ‘colour’ is North-American.
- To purge their systems of biases as well as ethically reprehensible answers, the companies behind these tools recruited people from all over the world to give feedback on the generated answers. The working conditions of these people are highly unclear, a fact that has been denounced in multiple media reports.
- Another ethical implication of GenAI is its impact on academic integrity. Frequently, the tools fail to include references to the information they generate. This absence of citations means that users must figure out who the original author was. It is the responsibility of the students and the researchers using these tools to ensure the quality of their research. Among other things, this means not presenting false findings and being transparent about the providence of their references.
- At first sight, it may seem that generative AI has the potential to eradicate inequality: all students have access to the tools, which means that e.g. relying on a private tutor for written assignments is no longer a prerogative of the rich. However, the developers of these tools are increasingly inclined to launch paying versions of their tool, which perform better/more efficiently than the free versions. If anything, this enhances inequality.
- Another aspect that must not be underestimated, is the ecological footprint of GenAI tools. The development and continued use of the tools require enormous quantities of computing power. The data centres where the tools are being trained and data is being stored cause a huge surge in power and water usage, the water being used for cooling purposes.
- Finally, there is the risk of anthropomorphism: the computer seems to think and talk like a human being, which may engender an unwarranted feeling of trust in the underlying systems. There is the additional risk that we may come to think of these systems as actual human beings, which in turn, may result in a decline in real human interaction. It is important to understand that the tools may have learnt to use several thought patterns from texts, but they contain no explicit logic and are very limited in their reasoning.
II. Gen AI Policy for Education at Ghent University
The easy access to AI systems forces study programme management and lecturers to think about the implications on their intended competencies/learning outcomes, on teaching and learning activities, and assessment. What is more, the professional field and society at large demand that we teach our students to use GenAI tools appropriately.
In close consultation with our Directors of Studies and/or our faculties’ education support staff, we have drawn up guidelines for GenAI in education for the 20204-2025 academic year.
Ghent University
- ... opts for responsible use of GenAI tools in the teaching practice, with a special focus on
- the validity of the assessment;
- the ethical implications;
- its impact on the student’s learning;
- preparing our students for the professional field and society with GenAI.
- ... chooses to explicitly allow the responsible use of GenAI tools in the context of the Master’s dissertation from the 2024-2025 academic year onwards;
- ... chooses to encourage the responsible use of GenAI tools in other (written) assignments from the 2024-2025 academic year onwards, and only to ban it if doing so is feasible and necessary for the assessment of the competencies/learning outcomes.
Validity of Assessment: Approach
In the 2023-2024 academic year, all faculties, study programmes and lecturers were asked to review the learning outcomes and assessment methods not only of the Master’s dissertation (course sheet and guidelines included) but also every other (written) assignment in an uncontrolled setting. The reason for this was that currently, Master’s dissertations and written assignments still start from the premise that students do not use GenAI tools. This means that the validity of the assessment is no longer at 100%.
Read up on how to review the Master's dissertation and other written assignments in these extensive guidelines.
Online written assessments call for extra vigilance as well. In the case of on-campus exams, physical surveillance is still highly recommended. In case of bring-your-own-device exams, there is no way to guarantee that students will not use GenAI. Shutting down internet access is not an option, either: any student with a modicum of IT skills can find alternative ways to go online. Holding online exams in computer classrooms on Ghent University computers is a safer option thanks to NetSupport School. Online off-campus written assessments are a challenge to organise and therefore we recommend them only for specific target groups.
An Approach to Fostering AI Literacy
Education has a major role to play in fostering AI literacy. Think, for instance, of the ability to use AI responsibly, or the ability to reflect on the use of AI, to interact with it correctly, and to be aware of its risks and limitations.
Study programmes, in other words, should offer teaching and learning activities that foster and assess students’ AI literacy. Since we allow the use of GenAI for Master’s dissertations and other (written) assignments at Ghent University, we give our students a chance to learn to use the tools responsibly. Foster AI literacy throughout the curriculum and identify course units that suit the purpose.
Make sure you students know what you expect in terms of AI use. If necessary, have them indicate in (writing) tasks how they used the tools so that you have visibility into whether or not they used the tools in a responsible way. You can use this template for this communication.
Looking for inspiration to foster your students' AI literacy? Point them to the info site on Ufora: Generative AI: from Concepts to Creation or implement the information in your course unit.
This implementation of course urges lecturers to be(come) well-versed in AI, too. There is no need to be an AI expert, but having basic skills comes in handy. The faculty and university education support staff offer workshops on AI literacy for lecturers. Ask about it at your faculty.
Looking to hone your own AI skills? Familiarise yourself with GenAI going through the Ufora info site Generatieve AI: over leren, creëren en doceren (in Dutch).
An Approach to Mitigating the Impact on the Curriculum
In the 2024-2025 academic year, Programme Committees will start charting the impact of GenAI on the programme- and course-specific competencies/learning outcomes, the teaching activities, and assessments throughout the curriculum. They can count on the support of the university education support staff, who have, among other initiatives, a specific offer for Programme Committees.
In these GenAI times it will be key:
- to reflect critically on programme competencies/learning outcomes, to determine which basic competencies/learning outcomes students have to acquire without (the help of) AI, to review and adjust learning outcomes, teaching activities and learning materials, and on how validity of the assessments can be guaranteed;
- to invest in digital literacy (including AI and GenAI literacy) of lecturers and students throughout the study programme.
Want to Know More?
This Education Tip is the result of consultations among Ghent University’s AI experts and educationalists, and based on information from the references below. If you have any further (support) questions, please get in touch with onderwijsondersteuning@ugent.be.
Sources
Adams, J., Brophy, L., Ediger, J., Herry, L. & Zumpano N. (2022). ChatGPT through an education lens. https://docs.google.com/presentation/d/1WeORhcE2tFOjI92MEMdYZK4wdBHVFOnVzrcc6rj1Pio/mobilepresent?slide=id.ga778454a28_0_111
Anseel, F. (2022, 8 december) De meest onderschatte vaardigheid. De Tijd. https://www.tijd.be/opinie/column/de-meest-onderschatte-vaardigheid/10432949.html
Cardona, M. A., Rodríguez, R. J., & Ishmael, K. (2023). Artificial Intelligence and the Future of Teaching and Learning Insights and Recommendations. https://tech.ed.gov
Clark, D. (2022) Donald Clark Plan B. http://donaldclarkplanb.blogspot.com/
Goethals, P. (2023, 24 januari) De grootste intellectuele hold-up uit de geschiedenis. De Standaard. https://www.standaard.be/cnt/dmf20230123_97406419
Liu, D. (2023, 11 mei). Care and Connection: Assessment in the Age of AI and Analytics. JISC Connect more
Miller, M. (2022). Ditch that textbook. https://ditchthattextbook.com/ai/#t-1671292150912
Molenaar, I. (2022). “Towards hybrid human-AI learning technologies.” European Journal of Education, 00, 1–14. https://doi.org/10.1111/ejed.12527
Monash University “Generative AI and assessment” 21 augustus 2023 https://www.monash.edu/learning-teaching/TeachHQ/Teaching-practices/artificial-intelligence/generative-ai-and-assessment#tabs__3173682-02
Rubens, W. (2022) Blog over ChatGPT. https://www.te-learning.nl/blog/
SURF (12 januari 2023) Impact ChatGPT op onderwijs [webinar]
UNESCO (2023) Guidance for generative AI in education and research https://www.unesco.org/en/articles/guidance-generative-ai-education-and-research
Van Deyzen, B (2023) ChatGPT-verzameling bronnen. https://communities.surf.nl/ai-in-education/artikel/chatgpt-verzameling-bronnen
Van Gorp, S. (2023, 10 januari). Moedig studenten aan de intelligente software ChatGPT te gebruiken. De Tijd. https://www.tijd.be/opinie/algemeen/moedig-studenten-aan-chatgpt-te-gebruiken/10439226.html
VU Amsterdam (2022). Hoe ga je als docent om met ChatGPT? https://vu.nl/nl/medewerker/didactiek/hoe-ga-je-als-docent-om-met-chatgpt
Watkins, Ryan (2022). Update your course syllabus for ChatGPT. https://medium.com/@rwatkins_7167/updating-your-course-syllabus-for-chatgpt-965f4b57b003
Appendix
- Experimentation Guidelines
- Checklist for Transparent Usage
- Step-by-step roadmaps for reviewing
- The student webpage: https://www.ugent.be/student/en/study-support/chatgpt
- Online learning path on GenAI for students: basic knowledge and use
- Online learning path on GenAI for lecturers: basic knowledge and use
Last modified Nov. 20, 2024, 8:24 a.m.