From high school principals' call for mandatory technological literacy to critical perspectives on AI image manipulation and predictive algorithms. Get insights into the latest developments in artificial intelligence in the education sector.
Here is this week’s newsletter on artificial intelligence in education.
We are in the process of rearranging things a little on Viden.AI to make the content more accessible. Right now, we feel that some of the themes we believe deserve a bit more attention — such as sustainability, ethics and disciplinary grief — are too far down the page. At the same time, we also want to make the site more lively and will therefore introduce a news section where we will continuously post news about AI in education. But don’t worry — we will keep our newsletter as a fixed, weekly summary of what has happened.
As it has been the Easter holidays, there hasn’t been much written about artificial intelligence in education in a Danish context. Therefore, in this newsletter we turn our attention to the United States to gain a more global perspective on trends we may also see in Denmark.
But before we get that far, we have three pointers from our own world:
While we await the forthcoming guidance from Styrelsen for IT og Læring (the Danish Agency for IT and Learning, STIL) on the use of generative AI in teaching, we turn our gaze to Norway.
While we have gained some valuable legal perspectives on AI and personal data from Norway, there is also relevant inspiration from the business sector.
In the United States we are seeing a number of challenges that we are likely already seeing unfold in Denmark, and which are therefore relevant to follow.
In addition, we have this week’s other news.
Today we are looking forward to taking part in the LEX conference on trustworthy knowledge in the AI age. Read more below, and sign up to get access to the video stream from the conference. It is at 15:00–17:30.
One of the things they will certainly talk about is the new collaboration in which Lex is teaming up with Aarhus Universitet (Aarhus University) to develop a Danish chatbot that enables users to engage with Lex’ content — without giving up personal data.
Photo: Liselotte Sabroe/Ritzau Scanpix
Happy reading.
Principal Martin Ingemann from Egaa Gymnasium criticises, in an opinion piece in Uddannelsesmonitor, the current direction of Danish education policy, which he believes ignores the fundamental changes in society driven by digital technology and AI. He points out that while the rest of the world is discussing advanced technology and digital tools, Denmark is still focusing on more traditional topics such as the literary canon and exam formats that do not reflect reality.
He urges us to ask the right questions of AI and technology — not how to protect ourselves against it, but how it changes our understanding of subject expertise, knowledge and learning. He argues that the education system should be an active part of society’s technological development and prepare young people for a reality that is changing rapidly.
Read the full opinion piece here:
Politiken has revealed, in an article, how OpenAI’s freely available image generator in ChatGPT has made it possible in record time for millions of users to create manipulated images. In just one week, 700 million images were generated. Although many use the feature for entertainment and creative play, researchers and experts warn of serious consequences for our understanding of reality and democracy. Experts such as Gry Hasselbalch and Yevgeniy Golovchenko point out that easy access to AI tools blurs the line between truth and fiction — which can undermine trust, spread misinformation and harm democratic dialogue.
The problem does not lie only in the technology, but in its massive availability. Manipulated images and videos can be used to undermine the credibility of individuals, create mistrust and promote disinformation. According to IT security expert Keld Norman, it is now easy for anyone, including criminals, to fabricate convincing image forgeries. OpenAI has assured that safety measures are built in, but critics believe the tool has already removed key filters and control mechanisms.
Professor Morten Mørup from DTU (Technical University of Denmark) highlights that we risk losing a shared understanding of reality because large-scale AI manipulation makes it difficult to distinguish truth from falsehood. This is especially problematic in a media environment where algorithmically curated information reinforces filter bubbles. The experts stress that source criticism and ethical considerations are now more important than ever, and that it should be standard practice to declare AI-generated material — especially when it involves real people.
In Berlingske, several prominent economists question the high expectations for AI’s impact on the labour market and productivity growth.
Professor Jesper Rangvid from CBS (Copenhagen Business School) is sceptical about the grand promises often attached to AI, comparing the current hype with the expectations surrounding the internet during the dot-com bubble in the 1990s. He points out that earlier technological breakthroughs — such as computers and the internet — have not led to measurable jumps in productivity, and that a similar pattern may repeat with AI.
Danske Bank’s Chief Economist, Las Olsen, simultaneously underlines that productivity improvements are difficult to measure in sectors where no physical goods are produced.
Professor of economics at Aarhus Universitet (Aarhus University), Philipp Schröder, supports this view and believes it is important to distinguish between short-term and long-term effects. Once the technology has been implemented, productivity growth will return to its original level. However, he emphasises that AI can contribute to increased efficiency in sectors such as the legal profession and education, where it has previously been difficult to achieve productivity gains.
Natasha Friis Saxberg from IT-Branchen (the Danish ICT Industry Association) also points out that AI is still a young technology, and that its long-term potential may be underestimated – even if, in the short term, it fails to live up to expectations.
Frederik M. Juel
In a longer opinion piece, Odin Hetland Nøsen, school consultant at Randbergskolen (Randberg School), argues that the understanding of when something constitutes personal data is often far too absolute – especially in the context of AI and education.
Drawing on both his own examples and recent EU court rulings, he shows that a piece of information only constitutes personal data when it can be linked to an identifiable natural person – and that this always depends on the context. When a recipient, for example OpenAI via an API, does not have access to identifying data or context, the information will, in practice, be anonymised for that third party.
Although the case-law of the Court of Justice of the European Union clearly points towards relative anonymisation, there remains a degree of legal uncertainty in this area. Institutions should therefore continue to assess each situation on a case-by-case basis.
Odin
A new book by Carsten Lund Pedersen and Thomas Ritter focuses on how AI projects can be implemented in a considered way so that they benefit the organisation. The book presents a range of models that provide structured, accessible guidance on implementing AI projects so that they create real business value. It also includes reflections on when an AI project should be abandoned.
We have reviewed the book, and it is aimed primarily at managers, IT specialists and consultants in the business sector.
Although the book is primarily aimed at business, we have chosen to feature it here. This is because it presents a number of models, many of which can be transferred to education in connection with the strategic implementation of AI in schools.
Carsten Lund Pedersen
In Time Magazine, Catherine Goetze describes how students’ use of AI is not simply about laziness or cheating, but rather a symptom of a broader crisis: our collective attention is under pressure.
She writes that even diligent and ambitious students feel mentally exhausted and are tempted to use AI because cognitively demanding work feels overwhelming. According to Catherine Goetze, the cause is not AI itself, but the constant influence of social media, which has accustomed young people to instant rewards and weakened their ability to sustain deep focus.
When AI is used as a shortcut, core learning processes are skipped, which undermines the ability to structure thoughts, build arguments and tackle difficult tasks. This not only weakens academic development but also creates a sense of inferiority.
But there are also many students who genuinely use AI constructively, for example to ask curious questions, have difficult topics explained, or create their own projects. Here, AI becomes a way to delve into subjects that are not always covered by the curriculum.
She argues that the solution is not to ban AI in school but to help pupils rebuild their capacity for deep work and intrinsic motivation. Teachers should meet pupils’ curiosity with open responses and use AI as a shared platform for inquiry – not as a surveillance tool for control.
She concludes by pointing to the need to hold the tech giants to account for the way their platforms have fragmented young people’s attention and created the conditions that AI is now criticised for exacerbating.
Catherine Goetze
The New York Times reports that many teachers are increasingly using generative AI in their work – while at the same time trying to limit pupils’ use of the same technology. Teachers use AI to plan lessons, mark assignments and analyse pupil data, which saves time and eases the workload. But this duality creates ethical dilemmas, for example when teachers ban AI-generated pupil work but themselves use AI for assessment.
The article highlights the growing industry around AI in education. Large investments and massive marketing by tech giants such as Google and Microsoft promise a digital transformation of school, but raise questions about the balance between automation and the human relationship at the heart of teaching. Many teachers express a desire for AI tools that support administrative tasks rather than replace the teaching itself.
Read the full article here:
At American universities, AI is increasingly being used to predict which students will complete their degrees. The aim is to target support, financial aid and guidance based on individual data such as grades, demographics and prior academic performance. But a new study shows that these algorithms may be biased, particularly against racial minorities such as Black and Latin American students. This is because the models are trained on historical data that reflect past inequalities in the US education system.
The analysis showed that Black and Latin American students were far more often incorrectly predicted not to complete their degrees—even though they actually did. These errors—so-called “false negatives”—occurred four times as often among Black students as among Asian students. This suggests that the algorithms overgeneralise from group-based patterns and fail to account for the variation that exists within racial groups. At the same time, the models often lack data on factors such as family support and study environment, which can greatly influence students’ wellbeing and performance.
Denisa Gándara
CTech
Jess Weatherbed
Matthias Bastian
Anders Lisdorf
Stephanie Convery
