AI hasn't just lowered the bar in schools. It has removed it entirely. Read this week's newsletter on the balance between AI as cognitive support and the danger of outsourcing thinking to machines.
Human beings are biologically predisposed to choose the easiest route, and that has always been the case. But AI has not merely lowered the bar in school. It has removed it altogether, and now we just walk through where we once had to jump.
This is the starting point for this week’s newsletter, with the balance between AI as cognitive support and the risk of cognitive impairment at its centre.
We start in the USA, where Professor Prince Sarpong argues that AI should support teachers’ assessment work, because our cognitive capacity is limited as the pile of assignments grows. At the same time, researcher Misia Temler warns that it has become too tempting to let AI handle tasks that genuinely require independent thinking.
These are two articles that ask the right questions, even though we still lack the answers. This is especially true when we must ensure that students do not end up losing the ability to think for themselves.
Here at home, journalism student Emil Probst writes about how AI can be a fantastic sparring partner, but also a dangerous shortcut in a pressured student life.
School leaders, students and DI (the Confederation of Danish Industry) are, in a joint appeal, calling for a national direction for the Danish primary and lower secondary school (folkeskole). In Billund Kommune they have not waited, but have already rolled out a concrete AI strategy to ensure that the technology is used in a pedagogically well-considered way and not as a shortcut to ready-made answers.
These are two perspectives that complement each other and set the framework for a necessary conversation about leadership.
The week’s other opinion pieces point to precisely the same issue.
The message is the same. We cannot leave it to the individual student or teacher to find the balance. It requires pedagogy, leadership and a shared direction if AI is to support learning rather than replace it.
At the same time, the technology constantly reveals new facets. A good example is researchers from Københavns Universitet (University of Copenhagen), who have used AI to analyse 850 historical novels, thereby expanding our cognitive horizon by identifying previously overlooked female trends.
It is precisely this balance between extending our capacity and outsourcing our thinking that I take up in this week’s commentary on taking the path of least resistance.
Enjoy reading the newsletter.
In the US magazine Times Higher Education, Professor Prince Sarpong writes in an opinion piece that AI should not be seen as a substitute for lecturers’ professional judgement, but as a tool that can support assessment and feedback.
He argues that the discussion about AI in higher education often gets stuck between enthusiasm and resistance, while an important point is overlooked, namely that humans’ cognitive capacity is limited. When instructors have to mark many assignments or theses, concentration wanes, and this can affect the quality, depth and consistency of assessment. Prince Sarpong therefore believes that AI should be used as support in assessment work rather than as a replacement for the lecturer’s academic judgement.
Read the full opinion piece here:
Prince Sarpong
In The Conversation UK, researcher Misia Temler writes that it has become increasingly tempting to let AI do tasks that previously required independent thinking. This includes, for example, writing, summarising and quick answers to questions.
She points out that research shows extensive use of AI can be associated with more superficial information processing, less critical engagement and a greater sense of dependence. Her point, however, is that the crucial thing is not whether we use AI, but how we use it.
Misia Temler explains that people have always relied on others’ knowledge, and that this in itself is not a problem. We learn precisely through support from others, for example when a teacher helps a student with feedback and guidance. But there is a difference between receiving support to develop one’s own understanding and handing over the thinking itself to others.
According to the article, AI should therefore be used as a tool that supports learning, instead of replacing it.
Read the full article here:
Misia Temler
In an opinion piece in Gymnasieskolen journalism student Emil Probst describes a clear double perspective on AI in the education sector. He highlights that AI tools can be a great support in study work because they can help to grasp difficult material, create an overview of extensive content and provide ideas for the writing process. In this way, AI becomes a practical sparring partner that can make the work more manageable for pupils and students.
At the same time, he warns that the growing use of AI may lead some to choose the quick fix rather than think, analyse and work their way through what is academically demanding. He stresses that this is not only about laziness, but also about heavy workloads, extensive syllabuses and a desire to get assignments done on time. The point, therefore, is that AI can both support learning and risk weakening it if the technology is used as a substitute for independent work.
Emil Probst
In an opinion piece in Skolemonitor, school leaders, students and DI urge a new government to set a clear national direction for the use of AI in the folkeskole. They point out that AI has already become part of many pupils’ schoolwork and teachers’ preparation, and that the question is therefore no longer whether the technology belongs in school, but how it can be used wisely and responsibly.
They therefore want a comprehensive and practice-oriented programme of trials, in which folkeskole schools systematically trial AI in teaching and in evaluation, so that the experiences can be used as a basis for political decisions.
Billund Kommune has spent more than a year and a half developing a concrete strategy for the use of AI in schools. The aim is to give teachers and students clear parameters, so that AI is used in a way that is academically relevant, pedagogically well thought through and with a focus on data and security. As part of the strategy, all teachers must take part in workshops, and the municipality has adopted a closed tool in which pupils can work with language models under more controlled conditions.
The strategy also emphasises that AI should be used to support idea development and learning rather than as a shortcut to ready-made answers. Students should be critical of sources, check the outputs, and be transparent about when they use AI. At the same time, experience from Billund shows that this work requires time, leadership support, and clear guidance on, among other things, GDPR, risks and professional development.
Billund’s experience is likewise that the work requires time, leadership support, and thorough clarification on, among other things, GDPR, risks and professional development.
In Uddannelsesmonitor’s election series, Moderaterne’s teaching and education spokesperson, Rasmus Lund-Nielsen, says that AI is not merely a new tool, but a technology that can take over important parts of the learning that education is supposed to give students.
He warns against students using AI to write assignments, analyse texts or solve problems in ways that weaken their own thinking and judgement. He therefore wants clear frameworks for the use of AI, more teaching and more tests with pen and paper, as well as a stronger focus on basic reading, writing and numeracy skills.
Taken together, the interview paints a picture of a party that wants more practice-oriented teaching, less central control and a more cautious approach to AI in the education sector.
Associate professor, PhD, and head of research on the teacher education programme at UCN (University College of Northern Denmark), Preben Olund Kirkegaard, argues that AI is not a problem in itself.
The technology can act as a kind of constant feedback partner, helping students explain concepts, structure assignments and develop ideas. But learning only happens when students themselves process and understand the suggestions that AI provides. It is therefore crucial that teaching focuses on students’ reflection and subject understanding rather than just the finished product.
Preben Olund Kirkegaard believes that teaching should therefore develop a genuine pedagogy for the use of AI. Teachers should design tasks in which reflection and academic judgement are central, and where AI becomes an explicit topic in teaching.
Students should learn how AI works, what limitations the technology has, and how to distinguish between a linguistically persuasive answer and an academically correct one.
In the podcast Undervisning & Formidling Anders Peter and Jeppe Bundsgaard tackle the question of what a good teaching resource should actually be able to do, and how teaching is affected by the materials teachers choose and use. The conversation moves from traditional teaching resources and subject-specific pedagogy to AI, motivation, community and practice-oriented learning.
Among other things, the podcast explores how teaching can become more inquiry-based, practice-oriented and collaborative, and what that means for how we view both teaching resources and technology. It also raises the question of what role AI should play in school, and how the technology can best support the teacher’s pedagogical work.
This is a conversation for those who want to delve deeper into the relationship between teaching resources, teaching and AI in school.
Anders Peter Nielsen
Publishing editor Maria Appel Bangshøj from Dafolo points out in Skoleliv that AI is taking up more and more space in school, but the work with the technology is organised very differently from school to school. Some schools have clear frameworks and a shared direction, while others leave the responsibility to the individual teacher.
She compares the situation with the introduction of learning platforms in 2016, when many schools used so-called superusers to help their colleagues. Experience showed, however, that it was difficult to create genuine ownership among all staff, because the change was complex and required a lot of time.
She therefore highlights a dual pedagogical responsibility in the work with AI. Leadership must create clear frameworks and a clear direction for the work, so teachers are not left alone with the task. At the same time, teachers must both use AI academically in teaching and help students understand when and how it makes sense to use the technology.
The point is that learning does not arise from the technology itself, but from the way teaching is planned and carried out. AI in school should therefore first and foremost be seen as a leadership responsibility and a shared pedagogical task.
The network Tech og Trivsel was formed in 2022 to focus on screen use, the influence of tech giants and the need for regulation.
It now points to five new challenges linked to AI: digital sovereignty, security and preparedness, psychological wellbeing, sustainability and the risk of cognitive outsourcing.
They warn against dependence on foreign platforms, energy consumption from data centres and a development in which people hand over thinking and relational work to AI. The network therefore calls for a broad debate on regulation and the responsible use of AI.
Read the full op-ed here:
Researchers from Københavns Universitet (University of Copenhagen) have digitised and analysed 850 novels from the period 1870 to 1900 using algorithms and AI. The project shows that our image of the Modern Breakthrough has been shaped by a relatively small selection of well-known authors such as Herman Bang and Amalie Skram. When the researchers instead examine a much broader body of texts, other themes, voices and works emerge that were previously overlooked.
One of the most important findings is that female authors did not only depict women’s lives as marked by oppression, sorrow and powerlessness. That tendency is still present, but the analysis also points to a clear current in which women’s lives are portrayed more positively, including through strong friendships and religiosity.
The project also highlights how AI can change literary research by making it possible to work with very large volumes of text. The digital methods have helped the researchers identify forgotten novels and authors and to nuance the understanding of the period’s major themes such as gender, gender equality and religion.
Technology has become easy. It's now lightning-fast to write texts, summarise books and find information, and every time a new technology challenges what we know, our first instinct is to regulate our way out of the problem. But what if rules aren't the answer this time?
The question isn't whether pupils will use it. They will. The question is what it does to them, because AI hasn't just lowered the bar. It has removed it entirely.
Think of a pupil sitting with an assignment at 11 p.m. Previously she might have opened her notes, leafed through the book and tried to formulate a sentence. She would have failed and tried again. It was laborious, and that was the point. Now she opens ChatGPT. In 30 seconds she has an answer that sounds right. She submits it. She doesn't fail, but she doesn't learn anything either.
It isn't cheating. It is a rational response to the fence having been removed. And the uncomfortable question is not "Why is she cheating?" The real question is "What have we designed that makes it rational to walk through instead of jump over?"
We tend to talk about friction as something to be eliminated at all costs. Some friction should indeed be eliminated. As Professor Prince Sarpong points out in this newsletter, it is good that AI can remove the unproductive friction for teachers who are drowning in marking. But there is another kind of friction. It is the kind that forces you to think, the difficulty that is the learning process itself, and the resistance that makes something stick with you as a human being. That is productive friction, and as researcher Misia Temler warns, it disappears when everything moves too fast.
In his book Menneskeetik, Svend Brinkmann describes the concept of "moral attention". It is the ability to notice that anything is at stake at all. It is not about what you do, or whether you make the right decision, but about whether you see the situation as it is in the first place. That is the hardest step in ethical action, and it is precisely what technology makes difficult, because it removes the pause, the little space and the three seconds in which you would otherwise have noticed it. Productive friction and moral attention are two sides of the same coin. The friction creates the space, and the space gives you time to see.
The solution is not to raise the fence with more AI bans or registration systems. We should not be building fences. We should create situations in which pupils themselves realise that they would otherwise have jumped over. We should cultivate friction deliberately, not as punishment, but as a method.
We should not rebuild the fence for the fence's sake. We need to teach pupils and ourselves to see when there ought to be one. That is digital literacy. It is not an IT project, but a human project. The question is no longer whether we jump over. The question is whether we even notice that there was a fence.
