Teaching and learning with AI: KNIGHT final conference

Under the guiding theme of "Artificial intelligence in university teaching", the two-day final conference focussed on the question of how AI can contribute to supporting teaching activities and supporting and evaluating learning processes.

Day 1: Keynotes, project presentations and impulses from practice

The event opened with welcoming speeches. Monika Renninger, pastor and director of the Protestant Education Centre Hospitalhof Stuttgart, welcomed the participants and emphasised the question of responsibility for AI-generated content. In her opinion, it is particularly important to ask critical questions when using AI. Prof. Dr Katja Rade, Rector of Stuttgart University of Applied Sciences, emphasised the global dimension of AI development in her welcoming address. Universities have the task of preparing students for a world full of complex challenges. This requires interdisciplinary collaboration, networked thinking and the ability to reflect critically. Projects such as KNIGHT enable the trialling of new teaching formats and contribute to the sustainable anchoring of AI-supported teaching activities at universities in order to improve teaching every day.

Prof Dr Judith Simon, Professor of Ethics in Information Technology at the University of Hamburg and Deputy Chair of the German Ethics Council, was invited to give the first keynote speech entitled "Teaching and learning with AI? Ethical and epistemological aspects" via video. She has been a member of the German Ethics Council since 2018 and deals with the intertwining of ethical, epistemological and political issues in the context of artificial intelligence and digitalisation in general.

In her keynote speech, she took a critical look at the ethical and epistemological challenges posed by the use of generative AI in university teaching, addressing the ten cross-cutting issues and recommendations of the German Ethics Council's statement "Man and Machine - Challenges posed by Artificial Intelligence". Using examples from university practice, it became clear how complex the ethical issues can be in the context of AI in teaching and that education must be understood as personal development.

This was followed by the keynote speech "(Artificial) Intelligence and the Great Transformation" by Prof Dr Jörg Kopecz, Professor of Corporate Management and Digital Transformation Management at the FOM University of Applied Sciences for Economics and Management in Bonn, Managing Partner of iTM - Institute for Transformation Management and Board Member of the Arbeitskreis Evangelischer Unternehmer AEU e.V. (Association of Protestant Entrepreneurs).

In his presentation, he posed the question of whether it is possible to develop education and society in such a way that they can remain sustainable with - or despite - AI. Using a SWOT analysis, he outlined Europe's role in the context of 'AI and transformation'. Mr Kopecz advocated a "fail fast" approach, more courage to change, agile processes and greater integration of interdisciplinary collaboration. His conclusion: the "train" has not yet left the station. In his opinion, expertise and mindset remain the key; future generations must be prepared for AI technologies and continue to receive fundamental training in order to maintain and develop the necessary skills.

KNIGHT work packages in detail and further keynote speeches

In the afternoon, the focus was on the results of the individual work packages of the KNIGHT project. These were presented by the professors responsible for the project, Dr Dieter Uckelmann, Dr Ulrike Pado, Dr Peter Heusch and Dr Tobias Popovic. The contributions showed the diversity of the approaches developed - from technical infrastructure to didactic concepts and ethical guidelines.

Work package 1 was dedicated to the conceptualisation of a skills matrix that forms the foundation for structured teaching and learning in the context of artificial intelligence (AI). The matrix differentiates between professional, methodological, social and personal competences. Work package 2 focussed on the development of a basic ethical concept. As part of a university-wide participation process, the AI ethics guideline of HFT Stuttgart was created. Work package 3 focussed on the development of a technical learning analytics and AI infrastructure. Among other things, the architecture of the "Study Dean Copilot", an intelligent early warning and action system for programme directors, was presented. The acquired infrastructure and its use in various projects were presented at the final conference. In work package 4 on the design of adaptive test formats, the Moodle plugin MOUSE was presented, which enables an adaptive selection of exercises. Initial tests showed that students learn more specifically and efficiently when they receive immediate feedback and can work on individual weaknesses. Work package 5 was dedicated to subject-specific implementation in the area of programming. The tool AVERT (Authorship Verification and Evaluation through Responsive Testing), which combines plagiarism detection with interactive feedback, was presented. Socratic questioning techniques are used to check the authorship of programming tasks - with a focus on the students' understanding. This was supplemented by the presentation of further education programmes in work package 6 to promote AI skills among students and teachers. The programmes are bundled into four topic clusters (technical, legal, ethical, social), have been made available to all students and staff and have been in high demand since 2022. HFT Stuttgart sees itself as a social actor that not only qualifies students professionally, but also enables them to reflect critically. Work package 7 dealt with the evaluation and sustainable anchoring of the project results. This included the evaluation of the AI certificate as well as various measures in the context of curricular anchoring. Work package 8 analysed the use of Large Language Models (LLM) in writing seminars and programming exercises. Among other things, it analysed how students write and compare their own and AI-generated abstracts - with the aim of promoting writing skills, not replacing them.

A coffee break was followed by presentations by HFT professors on other AI-related topics in teaching. In his presentation "AI-Assisted Grading of Student Answers", Prof Dr Sebastian Speiser showed how AI can support the grading of free-text answers. Through targeted prompt adjustments, assessment criteria could be applied in a differentiated manner - for example, by awarding partial points for partially correct answers. The question of which task formats are particularly suitable for AI-supported assessment remains a central research topic. Prof. Dr Anselm Knebusch reported on the use of GPTs to individualise teaching from the field of engineering mathematics. In a highly structured learning environment with weekly e-assessments, it was shown that students learn more over the course of the semester - even if long-term sustainability is still being investigated. The aim is to develop a tutor board that provides AI-supported individualised feedback and relieves the burden on teachers. Posters on the individual work packages could be viewed and discussed during the breaks.

Day 2: Project work in detail - workshops in the digital space

The second day of the event took place online and offered participants the opportunity to deal specifically with central topics of the KNIGHT project in the following workshops:

1. L(AI)tplanken - Ethical aspects of AI in university teaching: Case studies were used to discuss the ethical guideline developed in the KNIGHT project. The aim was to sensitise participants to normative areas of tension in the use of AI systems.
    
2. Conditions for the success of learning analytics in higher education: Two empirical studies showed the conditions under which students accept learning analytics - with a view to individual and institutional factors.
    
3. AVERT - Interactive author verification of programming tasks: An LLM-based solution for establishing authorship and proof of understanding was presented and discussed.
    
4. MOUSE and ASYST - Adaptive exercises and free-text assessment: Participants were given an insight into two tools for adaptive task selection and automated assessment. A live demonstration rounded off the workshop.
    
5. Study Dean Cockpit - AI for programme directors: The workshop demonstrated the implementation of a co-pilot for programme directors and addressed challenges in the areas of data protection and system integration.
    
6. Academic writing and AI: The potentials and limitations of AI-supported writing in academic writing seminars were discussed in a moderated exchange of experiences.

In small groups and breakout sessions, an intensive exchange was facilitated in which content, tools and results could be experienced in a practical way. The event concluded with a joint discussion in which key findings were summarised and perspectives for further work with AI in university teaching were discussed.

KNIGHT: A project with vision

Since its inception, the KNIGHT project has developed and trialled innovative approaches to the integration of AI in university teaching. The final conference impressively demonstrated how diverse and practical AI can be used in teaching - and how important interdisciplinary dialogue is for shaping these developments responsibly. The feedback emphasised the great interest in practical solutions and the desire for further exchange on the future of AI in university teaching.

Many thanks to all speakers, contributors, participants and all those involved in the project - especially the project staff - for their commitment, the exciting insights and the open, constructive dialogue that significantly enriched the final conference of the KNIGHT project.