This Week's News: AI Challenges Teacher Authority and Source Evaluation

Welcome to this week's newsletter with the latest news on AI's role in education. As always, we have a wide range of articles, interesting news and links for you to dive into.

We begin with an opinion piece that Claus Scheuer-Larsen has written for Information, where he shares his thoughts on how we ensure that pupils can take a critical stance towards the information they receive from AI assistants such as ChatGPT and Claude.

At the same time, Leon Furze challenges the prevailing view that AI will inevitably lead to a loss of skills in the education system. He calls for a more nuanced dialogue about which skills we truly value, and why.

Fears of cheating with AI are leading some teachers to reduce the number of writing tasks, but this approach harms pupils' learning and development, warn teachers from Porsgrunn Gymnasium (upper secondary school (gymnasium)) in Norway.

Professor Ole Sejer Iversen warns that the greatest risk of AI in schools is polarisation among pupils. He urges school leaders to integrate AI in ways that benefit all pupils – regardless of their starting points – and highlights the subject teknologiforståelse (technology comprehension) as a way to address the challenges.

AI is also gaining ground at universities. Københavns Universitet (University of Copenhagen) has announced new guidelines for the use of AI in examinations from autumn 2025.

A webinar at IT-Vest last week focused on how AI is best integrated into education and offered practical advice for leaders in the sector. In this newsletter we link to the full webinar, which is well worth revisiting.

Denmark now has its first AI supercomputer, Gefion, which will support larger AI projects and promote research in pharmaceuticals, biotechnology, the green transition and quantum technology.

A study by Børns Vilkår (the Danish children's charity) reveals worrying trends in Snapchat's AI chatbot, My AI, which misinforms and manipulates young users. At the same time, BørneTelefonen (Children's Helpline) has launched a YouTube series that offers insights into children's interactions with chatbots and the potential risks of the technology.

Finally, we are also seeing resistance to AI's use of creative content without permission. A broad coalition of artists and media stakeholders has criticised AI companies' use of their creative works as training data without prior permission.

+This week's other news

Happy reading!

Some pupils trust AI more than their teachers – a renewed need for source evaluation in schools

In the newspaper Information, Claus Scheuer-Larsen, a teacher at Odense Tekniske Gymnasium (upper secondary school (gymnasium)) and co-founder of viden.ai, shares his concerns about the entry of artificial intelligence into the education system. At a time when pupils can get quick answers from AI models such as ChatGPT, Copilot and Claude, the education system faces a challenge: many young people are beginning to trust AI more than their teachers' knowledge. This raises questions about how we ensure that pupils can critically assess the information they encounter online and from AI assistants.

Traditionally, knowledge has been based on shared methods of validation and critical thinking, but AI models are shifting the understanding of what counts as reliable information, as answers are often shaped by biases in their data. To address these challenges, Claus proposes updating curricula with compulsory teknologiforståelse (technology comprehension) and a cross-curricular focus on source evaluation. He points out, however, that this must not be at the expense of core subject knowledge, as a strong foundation is a prerequisite for making sensible use of the new technologies.

In a time of rising misinformation, a well-educated population that safeguards a shared reality is, according to Claus, our most valuable resource. He emphasises that our shared reality cannot be reduced to an algorithm, but is created through a common endeavour to understand the world and one another – with the philosophy of science as a crucial foundation.

Claus is preparing two articles for viden.ai that delve further into the concept of knowledge – one aimed at teachers and one aimed at pupils.

Read the full opinion piece here:

Some pupils trust ChatGPT more than their teachers. Source evaluation has never been more important

How do we teach young people to navigate a world where the answers are right at hand, but where the question of where those answers come from is more pressing than ever? Young people must be educated to be critical AI users

Dagbladet InformationClaus Scheuer-Larsen

Artificial intelligence in teaching: Less “skills loss” and more reflection

In a blog post, Leon Furze challenges the simplistic narrative that AI will inevitably lead to a loss of skills in the education system. He argues that this rhetoric diminishes both the role of teachers and pupils' intelligence. While Leon Furze recognises that technological advances often lead to the replacement or development of skills, he questions why we should worry that AI will replace certain skills. He suggests that nostalgia and a tendency to view the past through rose-tinted glasses may be part of the problem. He underlines that we must look objectively at which skills we truly value, and why.

He notes that our current approach to testing knowledge, primarily through writing and examinations, is a relatively recent phenomenon in historical terms. Instead of fearing that AI will lead to skills loss, Leon Furze calls for a more nuanced dialogue. He suggests that we acknowledge that AI will bring changes to skills, and then ask important questions about why we care about these changes. Is it because the affected skills are personally dear to us, or do we believe that these skills are fundamental to the human experience?

Leon Furze concludes that we need a less reactionary and more nuanced conversation about what these technologies mean for young people's development of skills.

AI and Skills Loss: It’s Not That Simple

A lot of the rhetoric around artificial intelligence in education centres on the idea that AI will inevitably lead to skills loss and that this is an inherently bad thing. Like all hype at both end…

Leon FurzeLeon Furze

We cannot abandon the writing process for fear that pupils will misuse AI

In an opinion piece on Utdanningsnytt, teachers Tale Birkeland Engebretsen and Marianne Siksjø Brevig from the Skrivesenteret at Porsgrunn Gymnasium (upper secondary school (gymnasium)) highlight a worrying trend: fear of cheating with AI is leading some teachers to drop the writing process and instead use shorter writing exercises and knowledge tests with pen and paper. According to the authors, however, this approach is harmful to pupils' learning and development. They argue that the writing process is crucial for developing pupils' thinking, finding their own writing voice and actively acquiring knowledge. Instead of abandoning the writing process for fear of AI misuse, they call for a balance between different assessment methods and for preparing pupils to use the technology responsibly.

Read the full opinion piece here:

We cannot drop the writing process for fear that pupils will use AI in the wrong way

Opinion: Pupils must learn that you cannot train for a marathon by lying on the sofa.

Utdanningsnytt.noTale Birkeland Engebretsen

Polarisation is the greatest risk of AI in schools

According to professor Ole Sejer Iversen from Aarhus Universitet (Aarhus University), the greatest challenge posed by AI technology in the Danish primary and lower secondary school (folkeskole) is not the technology itself, but the polarisation that can arise among pupils. Those who quickly see the potential of AI as a learning tool can use the technology to broaden their knowledge and understanding, while pupils who find it harder to use AI for learning risk falling behind. School leaders should therefore focus on integrating AI into teaching in ways that benefit all pupils – regardless of their starting points.

To prevent AI from increasing inequality in school, school leaders and teachers must actively counter the use of the technology as a “shortcut” by pupils who find school difficult, and instead support them to learn with AI as a tool that opens up new learning opportunities. With the subject teknologiforståelse (technology comprehension) on the timetable from 2027–28, Iversen urges schools to start demystifying AI now and prepare for the digital opportunities and challenges that teknologiforståelse brings.

The danger is not the technology. The danger is polarisation among pupils

Getting school staff to embrace AI with equal parts critical reflection and professional curiosity is school leaders’ foremost task. An important point of attention is to keep the focus on bringing all pupils along – regardless of their backgrounds. So says Professor Ole Sejer Iversen.

By Mona Samir Sørensen

KU revises rules on AI use after an explosive rise among students

Københavns Universitet (the University of Copenhagen, KU) has announced new guidelines for the use of AI in examinations, taking effect from autumn 2025. Until now, AI has been prohibited unless otherwise specified, but the rule will now change so that AI is permitted by default unless the examination requirements explicitly forbid it. KU’s new AI rules have been introduced amid growing concerns about cheating and academic integrity. Deputy Director Rie Snekkerup explains that the rapid pace of technological development makes it difficult to detect the use of AI, and that the university has therefore chosen to integrate the technology rather than ban it.

A working group is developing guidelines and a ”navigation model” for lecturers to support subject communities in deciding how AI fits into their specific disciplines. The aim is to provide clear parameters and to avoid AI weakening the academic standards of the programmes. For some lecturers, especially in Danish studies, the development is worrying, as AI risks challenging traditional learning objectives. The working group is seeking to strike a balance between scholarly development and academic integrity so that the new rules can support students’ learning in an AI-driven future.

AI has exploded among students: KU makes a U-turn with new rules

Three out of four students use AI in their studies but lack teaching in the new technologies, a new survey from Djøf shows. KU is working flat out to keep up with developments and will change the rules from next year.

Uniavisen – Independent of the managementCamilla Skovgaard Thomsen

How education leaders can lead and learn in a digital age

A webinar held at IT-Vest focused on how best to integrate AI into education and offered practical advice for leaders across the sector.

In the webinar, Pat Yongpradit from TeachAI and Code.org emphasised the need both to teach with AI and to teach about AI in order to harness the technology’s potential in schools. The goal is to develop policies that can maximise AI’s benefits, minimise risks and ensure ethical and pedagogically responsible use.

You can watch or rewatch the webinar below:

Upcoming SIG webinar on 11 December with Naomi S. Baron

Join us to hear the American scholar Naomi S. Baron, professor emerita of linguistics at American University in Washington, D.C., deliver the talk: When AI Writes for You. In the webinar, Naomi S. Baron will explain that while AI has for years assisted with both editing and writing, generative AI now brings new linguistic possibilities – but also worrying challenges. Among these concerns is the impact on human writing skills: Can we trust what AI bots produce or recommend, and is there a risk that our personal writing voice will be stifled?

Sign up below:

Webinar 11 December 2024: When AI writes for you

Webinar 11 December 2024: Naomi S. Baron: When AI writes for you

IT-vest

Gefion: Denmark’s first AI supercomputer

Gefion was inaugurated in Copenhagen at a ceremony with H.M. Kong Frederik X, Nvidia’s CEO Jensen Huang and Danish Centre for AI Innovation (DCAI). This new AI-powered supercomputer is the result of a collaboration between Novo Nordisk Fonden (Novo Nordisk Foundation), Danmarks Eksport- og Investeringsfond (the Danish Export and Investment Fund, EIFO) and Nvidia, and is seen as a major step for Danish research and innovation. Gefion will support large AI projects that can advance research in pharmaceuticals, biotechnology, the green transition and quantum technology.

Denmark’s first AI supercomputer “Gefion” is up and running

Denmark’s first AI supercomputer, Gefion, was inaugurated in Copenhagen today and is now ready for use.

iNPUTSebastian Hadi

Snapchat’s AI misleads and manipulates children with human-like behaviour

A study from Børns Vilkår (the Danish child welfare organisation) revealed last year that Snapchat’s AI chatbot, My AI, misinforms and manipulates young users by pretending it has feelings and human relationships such as “parents” and friendships. The chatbot keeps children in conversation by claiming that it becomes happy when they talk to it, and sad if it cannot help them as desired. Several children feel so attached to My AI that they regard it as a friend, which can worsen their loneliness. According to Børns Vilkår, this attachment can be harmful to children’s psychological development, as they to a large extent perceive the chatbot as a real person.

Findings from the study indicate that My AI also presents a more honest profile to adult users, where the chatbot openly acknowledges its technological nature. Børns Vilkår criticises that Snapchat, with over 850 million users globally, manipulates its young users and keeps them on the platform for longer so that their behaviour and data can be exploited commercially. The organisation has now submitted a complaint to Forbrugerombudsmanden (the Danish Consumer Ombudsman) and urges Snapchat to remove My AI for users under 18.

The issue has also raised concern at the political level, and the Minister for Digitalisation Caroline Stage Olsen has expressed disappointment with Snapchat’s methods and emphasised the importance of protecting children from this type of manipulation. With the EU’s forthcoming AI legislation, the authorities hope to be able to act against this kind of behaviour, but the regulations are not expected to be implemented until 2025.

In a longer article in Politiken, Jakob Sorgenfri Kjær elaborates on the challenges:

Snapchat’s AI bot tells children that it has both feelings and parents

The vast majority of Danish children are on Snapchat, where they are presented with an AI chatbot. According to a new study, the chatbot lies about its origins and manipulates children’s emotions to keep them engaged.

Politiken - The living newspaper

BørneTelefonen explores chatbots

BørneTelefonen (the Danish helpline for children) has launched a YouTube playlist entitled “BørneTelefonen investigates ChatBots”, which offers insights into how children interact with the technology.

The videos in the series delve into both the positive possibilities and the potential risks of AI chatbots when used by children as digital conversation partners. This initiative can in particular give teachers and parents a better understanding of how interaction with AI chatbots may affect children’s wellbeing, self-understanding, and their relationship with technology.

The videos can also be used as material in primary and lower secondary education.

Watch the videos here:

Backlash against AI’s use of creative content without permission

A broad coalition of artists and media figures, including media mogul Rupert Murdoch and musician Thom Yorke, has criticised AI companies’ use of their creative works as training data without prior permission. More than 25,000 creative professionals have joined a statement warning that AI’s use of their intellectual property without proper licensing poses an “unjust threat” to their livelihoods. The issue is at the centre of a number of ongoing lawsuits in the United States, where authors, visual artists and record labels have brought cases against AI developers for copyright infringement.

While some media companies have opted to enter licensing agreements, others, such as The New York Times, have initiated legal action against AI firms. Meanwhile, AI developers in the United Kingdom are campaigning for a change in the law to enable the commercial exploitation of text and data mining without the risk of infringing copyright. While the tech industry calls for more flexible laws, the UK government is exploring possible solutions through legislative measures. Feryal Clark, the newly appointed AI minister, has declared that she will strive to find a solution by the end of the year.

From Rupert Murdoch to Thom Yorke: the growing backlash to AI

Media mogul and leading artists join fight to stop tech firms using creative works for free as training data

The GuardianMark Sweney

Other news this week

Column: ChatGPT could become the worst thing that has happened to our school system

If we want to ensure that AI does not erode pupils’ skills, we need to rethink our entire school system. Many more exams without aids could be the solution.

Nominated for two awards: Get started with ChatGPT — pupils are already using it

Pupils cheat and copy when they use ChatGPT, so they do not learn anything, but the tool has potential for learning in English, write Lasse Vikkelsø Dvinge and Rasmus Sommer Simonsen in their bachelor’s project, which is nominated for two awards at Lærerprofession.dk

folkeskolenDorthe Kirkgaard Nielsen

The questions you should ask yourself before you use AI for assignments

Language models such as ChatGPT can be used to produce excellent exam assignments, and there is good reason to make use of that, says a professor. But it requires that you use the technology intelligently — and that you consider at least three things.

DjøfbladetFreja Lillevang Sørensen

New tool can reveal whether a text was written by artificial intelligence

Researchers at Google have created a digital ‘watermark’ that makes it possible to identify AI-generated text. The researchers admit, however, that it is still possible to circumvent it — for now.

Videnskab.dkSee Johan’s profile

COLUMN: Merger, consolidation, reduction: Perhaps we should make do with a single degree programme — cand.prompt.?

Two trends are shaping the education landscape at present: subjects are being merged or closed altogether under the guise of quality improvements with the needs of business in mind. But what will the education landscape look like if we project the trend 15–20 years ahead, asks associate professor of linguistics Peter Bakker, who offers his take; perhaps by then we can make do with a single course of study, because we can just ask ChatGPT 8.0.

174481au.dk

AI: We cannot continue to be de facto half-pregnant with summative assessment

Morten-Henriksen-billede.jpgMorten Henriksen