Priority areas

1. Re-design of examination formats

How do we develop examination formats that match a time in which AI is used as a tool by students and educators, and is increasingly and automatically embedded into the systems we use?

This creates new opportunities and expectations for how examinations are conducted, along with a range of dilemmas and major questions about what we are actually examining – and how it should be done. All programmes are already engaging with these questions, and there is a demand for joint initiatives and alignment across departments, as well as central support for developing new formats and enabling knowledge sharing across the organisation.

It may be worth considering designating one or more experimental spaces where we can test radical formats while still developing within existing frameworks.

2. StudyBots for students

An AI studybot must contain many layers of knowledge related to a student’s daily life and academic journey. This includes cross-disciplinary, general, and practical topics relevant to all students, as well as study‑specific and programme‑specific elements.

For example, we can experiment with a studybot that has access to all programme materials, the educator’s research, etc., so that an additional layer of support is developed in close dialogue with students and educators.

SDU should also consider using AI to support the study‑choice process. Other institutions have developed AI‑bots to assist prospective students based on study and course descriptions. This would be an obvious concrete initiative for SDU as well – otherwise, there is a real risk that young people will choose to study where they receive the most information and the best access to personalised communication during their study‑choice process.

3. AI in teaching development and delivery

How our programmes evolve in the near future will have a direct impact on our reputation. The university cannot continue with current teaching formats — AI as part of teaching and learning is not something that is on its way; it is already here.

There is a need for a prioritised focus on how AI can contribute to enhancing the quality of education, and for us to strategically ensure that students encounter a contemporary education with reflective and meaningful use of AI. Students are already using AI extensively, and many educators are actively integrating AI and redesigning their teaching.

There is a clear demand for prioritised central support to rethink teaching, including a pedagogical and didactic tech‑radar that keeps pace with developments and can provide recommended tools. At the same time, there is strong interest in facilitated knowledge sharing among educators who have already experimented with new formats, as well as among those who want to get started and seek peer‑based inspiration.

4. Server capacity and digital infrastructure to support research

For SDU to effectively support research in line with the rapid development in AI, a targeted investment and strategic enhancement across several areas is required. If SDU is to remain competitive in AI research, the infrastructure and support functions must be just as agile and modern as the research fields themselves.

There is enormous potential in using AI for data processing. Data processing that was previously too expensive has now become accessible in a completely different way. For SDU to support innovation with AI for handling complex datasets, including image analysis, a targeted investment in infrastructure, competencies, and resources is necessary.

Scalable computing capacity (GPU/HPC), secure and fast data storage, robust network infrastructure, as well as licenses and operation of AI platforms should be established. At the same time, it is crucial to ensure data security, compliance, and the development of interdisciplinary competencies. A focus in this area will also support SDU’s strengths and strategic initiatives within, for example, the drone field and the health sector, including the MedTech initiative. This effort can beneficially be aligned with one or more of these areas – not least in relation to external funding.

5. AI for research support processes – end-to-end

There is great potential in consolidating and integrating existing research support services—not organizationally, but communicatively. For example, combining SDU RIO’s tools, the library’s resources, and AI-based solutions into one complete package that can support researchers throughout the entire research and grant application process. This includes everything from idea development, literature searches, and interview transcription to data analysis, application processes, matchmaking with collaboration partners, and contract management.

Such a holistic approach can both increase efficiency and support the identification of new opportunities and new forms of collaboration across research fields. In many cases, there will also be valuable opportunities to optimize the functionalities of existing products.

6. AI to support meeting minutes and meeting flow

This initiative aims to explore and test how we can use AI to optimise workflows related to meeting activities. Some meetings are online, some are physical, and others are hybrid. All types of meetings can benefit from automated minute-taking functions in one form or another. The initiative should also explore possibilities for automating actions related to, for example, meeting format templates and meeting processes and workflows in general.

7. Maximised use of standard systems

Although this is an AI decision-making framework, there will, in connection with the streamlining of administrative processes, be solutions that need to be developed through the Power Platform or by activating process flows in major standard systems that are currently underutilised. In addition, AI functionalities are continuously being pushed into our existing systems, which we need to take better advantage of where it makes sense.

To create the greatest possible effect in relation to the optimisation of administrative processes, this is considered an area that should be explored early, with a view to assessing whether to implement additional AI agents or similar solutions. For example, Oracle, Microsoft, or Sitecore could be initial focus points, as this would generate efficiencies across the entire organisation.

8. AI to support scheduling and exam planning

There are many variables involved in timetabling and exam scheduling. AI agents can process vast amounts of data on room availability, number of students, lecturer preferences (e.g., preferred teaching times, teaching load), programme regulations, and previous patterns to generate the most efficient and optimal schedules. This can minimise downtime, avoid overlapping teaching, and maximise the use of resources. Once we gain control of these data, they can be used for exam scheduling as well.

9. Establishment of the SDU AI Hub

A clear pattern across the many interviews with the organisation is a strong desire for a central unit at SDU to support the development of our administration and core activities through AI – a technological and strategic AI knowledge hub that supports technological development, strategic coordination, and human bridge‑building around AI.

The unit should serve as a technological knowledge hub that continuously monitors developments in the AI field and tests new solutions. Overall, the AI Hub should be SDU’s proactive and agile gateway for all communication related to AI. It should support the organisation by providing access to knowledge about regulations, guidelines and security requirements, lists of approved and currently used systems, etc., while continuously mapping needs and requests.

This will include coordination of the newly proposed thematic Communities of Practice, which are cross‑organisational networks for knowledge sharing – a concrete and widely expressed wish across the interviews.

10. Prioritised focus on implementing targeted AI agents

There is significant potential in identifying administrative areas that, with strong data structures, can be supported by AI agents. This could include, for example, chatbots for internal guidance in areas such as finance, travel reimbursement, and GDPR. The aim is to create internal efficiencies through relatively uncomplicated agents that can quickly demonstrate their value and inspire the development of larger and more complex functionalities.

By prioritising development, rapid testing, and evaluation of AI solutions, new digital competences are gained, together with deeper organisational understanding of technological possibilities and limitations. This, in turn, can support the development of more advanced agent systems across IT platforms.

11. Access to language models

There is a strong desire within the organisation to gain access to language models beyond Microsoft’s Copilot – especially ChatGPT is widely requested. The available language models have different strengths and weaknesses and are therefore suited to different types of tasks.

The organisation needs to assess and decide which language models should be made available, and how to establish the appropriate security framework around their use. There are significant licensing costs associated with these software solutions, which must be considered in the decision, and an economic model will therefore be required for the use of different solutions.

12. Coordinated competence development – Academic Staff and Administrative/Technical Staff

There is a clear desire (and, in several cases, an expected requirement) for competence development for both academic staff and administrative/technical staff. This must be planned in a way that addresses the differing needs and motivations of the two staff groups while supporting the university’s strategic goals.

Academic staff typically have a high degree of autonomy and self‑direction. They do not want compulsory training programmes, but are motivated by academic development, knowledge sharing, and opportunities to contribute actively to their own and their colleagues’ learning environments. Researchers themselves highlight that this can take place in facilitated network structures – such as Communities of Practice. These may be supplemented with attractive Academic Masterclasses featuring leading researchers.

For administrative and technical staff, there may be a greater need for structured and compulsory competence development programmes that ensure up‑to‑date AI skills within administrative and technical functions.

13. Teaching spaces for experimentation with new formats

If teaching at SDU is to evolve in line with developments in AI, physical teaching spaces need to be radically rethought. We must be willing to experiment with physical spaces that incorporate entirely new technologies and create settings for dialogue in new formats.

It is proposed that three rooms be selected and designed to support three different types of interaction and learning. In addition to meeting lecturers’ wish for more flexible and alternative spaces, these three experimental physical rooms would provide a highly visible and tangible example of SDU’s active engagement with AI in education.

Coordination with the Epicur initiative on the establishment of hybrid teaching rooms will be required to ensure complementarity or potential integration.