AI as an everyday tool – and a shared responsibility

Dear colleagues,

Artificial intelligence is rapidly becoming an essential means to augmenting our capacities and underpinning our competences. AI can relieve us of menial tasks that take our time and hence eat into our capacity. Leaving such tasks to AI free us to do more demanding and rewarding tasks. And with AI we can access and leverage knowledge that without AI would take much longer to retrieve and were the forms of presentations might not in the first instances be tailored to how, with our individual backgrounds, we most rapidly comprehend and make knowledge actionable for us.

This might sound abstract, so let me exemplify. I was recently drafting an agreement that shall regulate how we will work together with a private company. Our standard collaboration agreements were useful points of departure but did not capture what was needed in this particular relation. My interlocutor in the company and I both had good ideas in non-legal terms of how we would like to work together. We could formulate that in layman terms infected by language we as engineers are comfortable with. Much might, however, get lost in translation when sending our draft to colleagues in the legal departments. Capturing our ideas in a language that would make sense to our legal departments was facilitated by AI.  Legal phrasings, that seemed opaque to me as an untrained person in matters of law and contracts, AI explained to me in terms I could comprehend. This brought to light needs for adjustments in the legal phrasings to achieve what the company and we wanted. This was an example of AI helping me access existing knowledge and make it actionable for me. The outcome, i.e. the draft collaboration agreement was one I fully understood and took responsibility for. I used AI but did not consider that any iota could be defended by stating that it came from or was recommended by AI. And I subsequently subjected the draft collaboration agreement to scrutiny and completion in our legal department, where, to my delight, it went through rapidly and smoothly.

Let me also share an example of AI relieving me of a simple menial job. For some automation, I was writing a piece of code with a nested set of loops. I initially created the set asking AI to complete a pattern. Subsequently I decided to invert the order of nesting for efficiency of the code, again simply asking AI to do the editing to invert. Easily done manually but a hundred times faster with AI. It would almost be a crime (misuse of TEK resources 😉) not to leave such things to AI.

At TEK, we are now at a point where we should wholeheartedly embrace and use AI in our daily work. We should do this across our administration, research, and education. In doing so we must see AI merely as a powerful tool and in no way as a replacement for human judgment or responsibility. Used responsibly and skilfully, AI will be an extraordinary amplifier of our capabilities and capacities.

AI in Administration: Better, Faster, and More Rewarding

In our administration, we should actively explore how AI can support the many tasks that keep our faculty running. Whether drafting minutes of meetings or letters of recommendation, analysing processes, preparing cases, or responding to complex queries, AI can help. Used well, it can make our work more efficient, improve the quality of what we produce, and free time for the parts of our work that require human insight and nuance.

But with this power comes responsibility. AI can assist us, but it cannot relieve us of responsibility. Every member of staff must always fully understand and take ownership of what they produce with AI as a tool. The human remains fully and solely accountable, not AI.

AI in Research: Standing on Shoulders

Research, by its nature, breaks new ground, goes where none went before and develops new methods. But in doing so, it stands on the shoulders of what is already known and makes use of methods already developed. Here, AI provides a unique opportunity. It is easier than ever to explore vast bodies of literature, identify relevant prior work, compare perspectives, and discover connections across fields. And it reduces the time used to access and operationalise existing knowledge. This accelerates the research process and strengthens our foundations for truly novel ideas, the ones only humans can generate.

Quality assurance is another dimension where AI can help us tremendously. For internal pre-submission review of a scientific manuscript, checking arguments for coherence, and ensuring consistency in methods, AI is a valuable sparring partner. The same applies to preparing grant applications. With AI, we can more quickly identify relevant calls, relate our competences and ambitions to the specific calls, situate our ideas within the state of the art, and strengthen the clarity and structure of our proposals. Where addressing a societal challenge is central to a call, AI can help us clarify relevance and impact of our proposed research, an aspect of our grant proposals where, according to reviews, we (on average) notoriously underperform.

As always, accountability sits fully and squarely with the researcher. AI is a powerful tool, but only a tool. We, the researcher, should never present scientific results which we are not fully on top of and vouch for. If the nail goes in crooked it is not the responsibility of the hammer but that of the carpenter.

Collaboration and Talent: Scouting Further

Research thrives through collaboration. Traditionally, collaboration emerges through conferences, networks, and reading the literature. AI expands this horizon significantly. It allows us to scan far more broadly – and across disciplines – when we look for potential partners. It can reveal groups, individuals, or emerging themes we might never have encountered otherwise.

The same applies to recruitment. When we have open positions, identifying excellent candidates is essential for TEK’s future. AI can help us identify and explore a wider talent pool and more rapidly examine a candidate’s profile. But, again, the final judgment must be ours.

AI in Education: Great Potential assuming Important Safeguards

In education, AI presents both a tremendous opportunity and a clear challenge. Used well, AI can support learning in new and highly effective ways. Students can explore multiple pedagogical explanations of the same concept, receive tailored challenges calibrated to their level, and test their understanding through AI-based dialogue. Educators can use AI to design course material more efficiently, create variations of assignments, and support students individually at scale.

But we must avoid the trap where AI for some students becomes a shortcut that undermines the rigor of learning. Students must still build a solid foundation in their disciplines. A foundation that enables them to use AI wisely rather than rely on it blindly. We must help them use AI in ways that deepen their understanding rather than replace it.

At the same time, we must prepare our students for a world where AI will be part of almost everything they do and where they will be out-paced if they are not AI-literate. Our graduates must leave TEK not only with disciplinary excellence but also as confident, responsible users of AI, capable of leveraging these tools without ever abdicating responsibility.

A Shared Path Forward

Across administration, research, and education, AI is becoming essential. Not to replaces us, but to enhances what we and only we can do: think, judge, imagine, and take responsibility.

My hope is that all of us at TEK will explore AI and embrace it as a daily tool. I encourage all to help and inspire colleagues and conversely reach out to colleagues for inspiration and help. Let us integrate learning on use of AI in our everyday and complement where needed with dedicated training. We will augment the broad effort with ambitious actions on selected areas supported by our in-house AI experts.

While much of our work does not involve confidential information, some does. For the latter we must only use AI in a form that does not feed learning to AI that can be accessed by the whole world. To facilitate that, SDU has a rudimentary version of Copilot in a cordoned off dataspace, that all staff can use. SDU is exploring a more powerful version of Copilot and initially making it available, at some cost, to selected staff. Ultimately, we will at TEK make such tools available to all with a commensurate need. There is a plethora of AI tools. Do explore and makes us collectively able to use those best suited to our many different tasks. Most come with a free version with limitations that become irritating when moving to larger jobs and goad us on to buy subscriptions. To help our learning we will accept such expenses in a reasonable measure. I prefer that when trying to make something work and I fail that it is not due to the inadequacy of the tool but my own shortcomings. Then I know what to work on.

AI will change the way we work. We need all to be proactive in this as domain knowledge is essential to identify how AI can assist. We will do this without concern about making ourselves redundant. At TEK we need all-hands-on-deck in this and complement it with the social contract that no one will be made redundant because of AI. TEK is growing, mainly in research and somewhat in education, and will continue to do so in the foreseeable future. There is much we would benefit from doing today that we cannot do with available resources. With further growth, demands will grow. So, all our highly capable and dedicated staff will be needed. Where tasks shift, for instance in our administration, we will ensure time and training to facilitate transition to whatever the new will be.

Used wisely, AI will make our work more efficient, our research stronger, our teaching richer, and our graduates better prepared for the world they will shape. Let us embrace these tools, not as shortcuts, but as companions in the work that remains unmistakably human.

All the best,

Henrik