This article has been machine-translated into English. Therefore, there may be nuances or errors in the content. The Danish version is always up-to-date and accurate. Read the original Danish article here.

This week’s news on artificial intelligence in education again focuses on challenges with cheating and the need to adapt exam formats.

The debate about AI in teaching and in exams has been simmering for a long time. We have written about it in many of our newsletters, but Henrik Nevers’ opinion piece in Berlingske appears to have been a catalyst for increased attention to the challenges. The Danish Minister for Children and Education was also interviewed in Berlingske, where he signalled that more exams may in future take place with pen and paper.

The debate is far from over, and it seems as if the whole sector is calling for action, which is particularly evident in the many opinion pieces and op-eds published last week.

Amid the heated debate, it is important to stay calm. While everyone is discussing solutions, we still need to answer the fundamental question: Which core competences should students be able to master independently before they can use AI meaningfully?

Special Consultant Ove Christensen has published an interesting post on LinkedIn, pointing out that it is the student’s practice and engagement that determine the learning outcome.

Center for IT i Undervisningen (Centre for IT in Education) has published a new study from the vocational education and training (VET) programmes, in which students ask for more teaching in technical application, ethics and concrete tools. The report is accessible via the link.

A completely different perspective is that AI is changing our language, and The Verge has published an interesting article about this. We are losing the small linguistic mistakes, the humour, the regional flavour and the language’s vulnerabilities – elements that usually affirm our humanity.

We have also compiled this week’s news from the tech world. Finally, you’ll find articles we didn’t have space to cover in more detail.

Enjoy the newsletter!


AI challenges the credibility of upper secondary school (gymnasium) students’ grades

In Berlingske, Henrik Nevers, principal at Roskilde Gymnasium (upper secondary school), writes that students’ assignments have improved linguistically, but that in many cases this is due to AI and not necessarily academic development on the part of the students. This gap between coursework and exam performance creates distrust of grades and thus also of the exam certificate.

Henrik Nevers therefore proposes that written annual grades in certain subjects should be abolished because they are no longer reliable. Instead, written and oral work could be combined into a single year grade.

Upper secondary schools warn: Artificial intelligence makes the exam certificate unreliable
Read more here.
Behind a paywall

Minister announces more pen-and-paper exams in response to AI use

From the next school year, certain upper secondary school students will sit exams exclusively with pen and paper. This applies, among others, to subjects such as French beginner’s A, Italian A and Spanish beginner’s A. The decision is based on experience from earlier trials showing that the analogue exam format strengthens students’ ability to produce language independently.

Mattias Tesfaye, the Danish Minister for Children and Education, sees this as a balanced strategy in which AI should be included in some subjects and exam formats, while others are completely decoupled from digital aids.

“We are starting with these language subjects, but I can easily imagine it in several other subjects,” says Mattias Tesfaye to Berlingske.

He emphasises that the ability to write clearly remains a core competence that must be preserved and strengthened.

Analogue comeback in upper secondary school (gymnasium): Minister announces more pen-and-paper exams
Read more here.
Behind a paywall

Upper secondary schools want to rethink exams and integrate AI

Speaking to DR, Maja Bødtcher-Hansen, chair of Danske Gymnasier (the Danish association of upper secondary schools), explains that AI creates distrust and uncertainty among both teachers and students. She proposes that AI should in future be integrated as part of the exam format, for example by allowing students to work with the tools and then document and reflect critically on their use.

Both Gymnasieskolernes Lærerforening (the Danish Union of Upper Secondary School Teachers, GL) and Center for Undervisningsmidler (Centre for Teaching Resources) support the proposal. They see the written exams in particular as challenged, because language models can generate answers that make it difficult to assess a student’s independent work.

In addition, a technological arms race is playing out between proctoring software and creative students trying to circumvent the systems during exams.

Cheating with ChatGPT haunts upper secondary schools: Now teachers and experts want to completely rethink exams
Artificial intelligence should be built into exam tasks, several parties believe. The minister is working on the issue.

Upper secondary school students demand that AI be a natural part of exams

Many upper secondary students use AI tools such as ChatGPT extensively in their studies but experience a mismatch when these tools are banned in exams. Students like Joachim Ekstrand Federspiel and Luna Lykke Sørensen describe it as both frustrating and unfair to be allowed to use AI all year but have to forgo it in the exam. According to the students, it is equivalent to removing the calculator in mathematics or fighting with your hands tied behind your back.

There is therefore also a wish for more systematic teaching in AI and technological literacy, preferably as a separate subject in upper secondary school.

A report from Danmarks Evalueringsinstitut (the Danish Evaluation Institute, EVA) shows, however, that AI creates uncertainty and distrust among teachers and external examiners because it is difficult to assess how much the student has contributed themselves. Many therefore point to a solution in which AI may be used, but where the exam is followed by an oral account and critical reflection on its use.

Students are heavy users of ChatGPT throughout the school year, but AI is off-limits in exams: ‘Equivalent to denying us a calculator or Excel’
Schools, students and the minister all believe exams should be rethought to fit an age of artificial intelligence.

Upper secondary schools must rethink exams and teaching with AI rather than return to pen and paper

Author Stefan Flensmark, upper secondary teacher and AI course instructor, argues that we should drop the discourse of cheating and instead see AI as an opportunity to rethink how we teach and conduct exams.

He believes this development has arrived at a pace that surpasses earlier technological advances, and that Danish upper secondary schools need to adapt.

According to Flensmark, it will be a bloodbath in the exam if we continue as we have done with purely traditional methods.

Instead, schools should open their doors and develop teaching and examinations that embed AI as a tool, both in the learning process and in documenting students’ academic competences.

I expect a bloodbath in the exams
In students’ reality now and in the future, AI tools will be ubiquitous. It is therefore crucial that upper secondary schools (gymnasiums) equip students to use them reflectively and critically. A return to pen and paper will not do.

Examinations and teaching should reflect responsible use of AI

In an opinion piece on Skolemonitor, principal Lars Roesen from Ringkjøbing Gymnasium (upper secondary school) argues that the education sector can no longer ignore AI’s presence and significance. He emphasises that the decision to use AI in society has already been made, and that both private and public actors are making extensive use of the technology.

Schools and upper secondary schools must therefore keep pace and take responsibility for teaching students to use AI in a critical and responsible way.

He warns against creating an examination practice that does not reflect the reality students encounter outside school. If examinations do not include the use of AI, we risk creating an artificial parallel world that does not prepare students for the digital reality.

In the opinion piece, he explains that AI should not be seen as a threat to academic standards, but as a technology that actually requires subject knowledge in order to be used responsibly and appropriately.

Principal: The fight for knowledge goes hand in hand with the responsible use of AI
We fail in our obligation to prepare our students if, in examinations, we create a parallel world that sets requirements which do not match those they will meet after and outside school.

AI requires a break with outdated assignments and teaching methods

In an op-ed in Avisen Danmark, Søren Rinnov Østergaard, country manager for Samsung Danmark, takes as his starting point his own son’s use of AI to write a Danish assignment in 15 minutes.

According to Søren Rinnov Østergaard, we are in the midst of a fundamental change that has yet to be met with a corresponding willingness to reform — neither from the political system nor from the education sector itself. He warns that essays and homework assignments will increasingly become meaningless if they are not adapted to a reality in which AI is readily available.

The son’s cheating with lektor Blomme was a clear sign: Now Samsung’s Danish chief issues a call to action
I actually did a double-take when my son in Year 7 had to read and analyse “Det forsømte forår” by Hans Scherfig. He had 14 days for it. Thirteen days passed before he began. Even so, he did the analysis and wrote the assignment — in just 15 minutes.

Learning with AI depends on engagement, not on the technology itself

The debate about whether generative AI strengthens or weakens learning is, according to senior consultant Ove Christensen, wrongly framed. The question is not whether one learns from AI, but how the technology is used.

A new study from Københavns Professionshøjskole (Copenhagen University College) shows that fewer than 5% of students avoid AI entirely, while the vast majority use tools such as ChatGPT for tasks like inspiration, understanding, translations, information retrieval and writing assistance. Writing assistance is an area where AI can both support learning and challenge academic authenticity.

The study points to two ways of using AI: either to outsource one’s study work, which reduces learning, or as a partner in an active and reflective process. It is the latter that supports academic development. The point is that the technology does not determine the learning outcome; the student’s practice and engagement do.

Can you learn anything with generative artificial intelligence?
Does generative artificial intelligence (GAI) help improve or damage students’ learning and academic development? Different studies point in different directions. A study from MIT, for example, has suggested that the use of GAI in the longer term undermines the ability to think, because a

The National Student Panel 2025 on AI in teaching

Center for IT i undervisningen (the Centre for IT in Education, CIU) has conducted a survey among 745 students from vocational education programmes. It shows that the use of AI has become widespread. 64% have used it in their studies, particularly for understanding, inspiration and linguistic support.

Many students express concern about AI’s impact on their ability to think for themselves, the risk of cheating and uncertainty about the rules. A quarter of the students have used AI to write entire assignments, and more than half believe that this is cheating. Uncertainty about when the use of AI is acceptable characterises many students, who also encounter varying attitudes among teachers.

The study points to a clear need: students are asking for teaching in technical application, ethics and specific tools.

7 out of 10 would like to learn more about AI – especially how the technology is used responsibly. There is a particular desire for a shared introduction and clear guidelines for the use of AI in teaching.

The Student Panel recommends three initiatives: that all students receive the same introduction to AI, that all teachers actively consider AI’s role in their subjects, and that teaching is planned so that AI supports students’ academic reflection rather than taking over their work.

https://videnscenterportalen.dk/ciu/katalog/elevpanel-ai/


AI is affecting our language use – and risks making us less human in conversation

The Verge has published an article in which researchers at the Max Planck Institute show that ChatGPT is influencing our language. Analyses of over 280,000 YouTube videos document a marked increase in the use of words preferred by AI.

It often happens without users noticing. AI’s linguistic habits creep in and shape both our speech and our thinking. But AI’s influence extends beyond word choice. Studies show that communication becomes more structured, emotionally neutral, and characterised by a more uniform AI tone.

This leads to a new form of mistrust: people perceive language that could have been generated by AI as less authentic, even when it is not. It is not the use of AI itself that arouses scepticism, but the suspicion of it.

Researchers warn about the loss of human signals: small linguistic slips, humour, regional character and vulnerability, which normally confirm that we are flesh-and-blood humans. These signals disappear when we let AI do the wording for us. This can weaken trust and identity in both private and public communication.

You sound like ChatGPT
AI isn’t just impacting how we write — it’s changing how we speak and interact with others. And there’s only more to come.
Empirical evidence of Large Language Model’s influence on human spoken communication
Artificial Intelligence (AI) agents now interact with billions of humans in natural language, thanks to advances in Large Language Models (LLMs) like ChatGPT. This raises the question of whether AI has the potential to shape a fundamental aspect of human culture: the way we speak. Recent analyses revealed that scientific publications already exhibit evidence of AI-specific language. But this evidence is inconclusive, since scientists may simply be using AI to copy-edit their writing. To explore whether AI has influenced human spoken communication, we transcribed and analyzed about 280,000 English-language videos of presentations, talks, and speeches from more than 20,000 YouTube channels of academic institutions. We find a significant shift in the trend of word usage specific to words distinctively associated with ChatGPT following its release. These findings provide the first empirical evidence that humans increasingly imitate LLMs in their spoken language. Our results raise societal and policy-relevant concerns about the potential of AI to unintentionally reduce linguistic diversity, or to be deliberately misused for mass manipulation. They also highlight the need for further investigation into the feedback loops between machine behavior and human culture.

AI news from the tech world

Here is a brief overview of new or improved AI technologies that appeared last week.

  • Midjourney has launched its first AI video model, called V1, which can animate still images into short video clips of 5 to 20 seconds.
    Source: Midjourney
  • OpenAI has launched “Record mode” for Pro, Enterprise and Edu users on macOS. The feature provides live transcription and summarisation of meetings and voice recordings of up to 120 minutes, as well as automatic, time-stamped actions.
    Source: OpenAI
  • Google has launched “Search Live” in AI Mode (Labs, USA), enabling voice- and conversation-based search via the Google app. The feature offers both speech input and audio output with Gemini AI integrated. Visual search is on the way.
    Source: Google
  • MiniMax-M1 is a new Chinese open-source language model with an enormous context window (over 1 million tokens and 456 billion parameters). The model also includes an AI agent, MiniMax Agent, that can write code, build apps and create presentations.
    Source: Fortune
  • Meta and Oakley have introduced their first Performance AI glasses: Oakley Meta HSTN with a built-in 3K camera, high-resolution Ultra HD video recording and improved battery life.
    Source: Bloomberg

This week’s other news

Can we still trust written end-of-year grades? “It depends…”
Researchers warn: One in four children use AI – but without protection
Nearly one in four children aged 8 to 12 already use generative AI technologies such as ChatGPT and Dall-E, according to groundbreaking new research from Alan
Teachers have become “cheat-checkers” – now a headteacher proposes scrapping end-of-year grades
The Minister for Children and Education, Mattias Tesfaye, rejects scrapping the written end-of-year grades.
Upper secondary school (gymnasium) students call for AI in exams
When thousands of upper secondary students across the country sit their exams in these weeks, they are not permitted to use artificial intelligence in the tests. This is despite AI being a major part of students’ everyday lives.
Are you heading back to the classroom? Here’s what one of the options offers
Many have already applied for the new funding for continuing education, but there is still plenty of support available. Folkeskolen.dk/engelsk has looked more closely at one of the courses offered nationwide.
🔍 AI and written exams – the conversation should start in the classroom, not at the exam desk: | Ole Sejer Iversen
🔍 AI and written exams – the conversation should start in the classroom, not at the exam desk: 💬 Here is a conversation starter for schools and upper secondary schools (gymnasiums): Understanding artificial intelligence in education is a marathon, not a quick fix. 👉 If we ban AI, we risk making education irrelevant. 👉 If we allow AI uncritically, we risk making students irrelevant. There are no easy solutions, but there are good questions. If we are to create meaningful integration of AI in exams and in teaching, we as leaders, teachers, IT managers and board members must take a stand – not just on the technology, but on the educational and formation task. 1️⃣ How does the school/upper secondary school (gymnasium) work with digital communities and technologies as part of the school’s/upper secondary school’s educational and formation task? 2️⃣ Do we see AI as a threat or an opportunity for students’ learning and development – and why (👀 see the conversation starter in the table below)? 3️⃣ Which specific AI-related goals and initiatives does the school/upper secondary school have for the next 1, 2 and 5 years? 4️⃣ Which frameworks and exam formats should we seek to influence politically and pedagogically – based on our values and experiences? Would this process work at your school or upper secondary school? Videnscenter for Digital Teknologiforståelse (Knowledge Centre for Digital Technology Comprehension), Afdeling for Digital Design & Informationsvidenskab (Department of Digital Design & Information Studies), AU, Faculty of Arts - Aarhus Universitet (Aarhus University), | 22 comments on LinkedIn

Share this post

Written by

Per Størup Lauridsen
MSc in IT | Co-founder of Viden.AI | Head of IT and Communications at TietgenSkolen