Like most countries around the world, the Nordic nations of Denmark, Sweden, Norway, Finland, and Iceland have been caught up in the recent ChatGPT hype, and their universities are actively grappling with how to respond and benefit from this technological leap. In Denmark, some universities have already banned the use of ChatGPT in examinations. Anna Bak Maigaard, Deputy Director at Aarhus University, told Danish broadcaster DR that exams must be completed individually and using artificial intelligence means the student is not working alone. The university announced this ban on January 3. Similarly, the Technical University of Denmark has officially prohibited AI use during exams. Professor Philip John Binning, dean of graduate studies and international affairs there, admitted the university currently lacks a clear strategy and described ChatGPT as “creating a new reality” that requires the institution to think actively about how to harness the technology effectively.
At the IT University of Copenhagen, Lene Rehder, Head of Student Affairs and Programmes, said no official ban on chatbots during exams has been issued, but their use is not allowed in “unaided” exams. She compared the current situation with the early days of the internet, noting that how technology is used depends on the nature of the exam. When exams do not permit any aids, ChatGPT is naturally prohibited.
However, opinions differ within Denmark. In February 2023, Hans Stokholm Kjer, a professional project manager and freelance commentator, argued in Akademikerbladet magazine that forbidding students from using ChatGPT even during exams is a defeatist attitude. He pointed out that while ChatGPT is often mistakenly called “intelligent,” it is essentially excellent at sorting and reproducing information it has been fed — and it will only get better. He emphasized the importance of embracing the technology fully to avoid educating graduates with outdated skills from the 1950s. Instead, he believes graduates should be prepared to navigate developments toward and beyond 2050. In this light, exams should not only measure students’ knowledge but also their ability to select and use the right tools, including AI, where appropriate.
At the same time, Hanne Leth Andersen, rector of Roskilde University and chair of the education policy committee at Universities Denmark, highlighted the need for clear and unambiguous rules governing AI use in university settings. She said the sudden leap AI took with ChatGPT puts pressure on universities to react swiftly, as the chatbot will affect both teaching and exams. Andersen stressed that policies must be closely tied to specific learning objectives of each course and programme. While teaching staff are still beginning to integrate these tools, it is clear that AI will be a powerful force in the future. Students must learn not only how to use AI but also when it is helpful and when it is not. In education, the goal is often not simply to find the right answer, but to track one’s methods and references accurately. Andersen also pointed out that adapting curricula and teaching methods to new technology is not unprecedented — universities have successfully done so before with computers, calculators, and the internet. She is confident the same will happen with AI, respecting academic quality as the foundation.
In Sweden, the technology has already presented practical challenges. At Uppsala University, a student received a warning for using ChatGPT during an exam, marking the country’s first disciplinary case of this kind. The disciplinary committee found the student had attempted to mislead the examination board by using ChatGPT to complete the test but issued only a warning since the AI-assisted tasks made up a small part of the exam. Professor Mikael Wiberg, affiliated with Chalmers University of Technology and Umeå University, confirmed that students have submitted papers referencing ChatGPT, though formal policies on AI use are still being developed. Professors across Sweden have noted the need to teach students how to approach AI tools responsibly, particularly in programming courses. Martin Duneld, a lecturer at Stockholm University, warned that the rapid advances in technology might make exams more expensive and called for a return to supervised, in-person exams rather than online formats to ensure learning objectives are met.
Some academics, like Anders Isaksson of Chalmers University, view ChatGPT as an opportunity to develop new skills around AI prompt engineering. Isaksson revealed that he used ChatGPT extensively to write a blog addressing concerns about AI in exams. He explained that while ChatGPT generates the text, the quality depends heavily on how users craft their prompts, making the ability to design effective prompts a crucial skill for future writers.
In Norway, universities have taken a proactive approach. Seminars were held in early February at the University of Bergen and the University of Agder to discuss ChatGPT’s impact on education and research. Professor Arild Raaheim of Bergen described the initial reactions as either overly dramatic, overly cautious, or constructive. He identified three common responses: paralysis or a “wait-and-see” approach, exemplified by the Norwegian University of Science and Technology proceeding with spring exams unchanged; overreaction focused on restrictions and fear, which often leads to missed opportunities; and a proactive strategy that combines new and traditional methods to strengthen student learning and critical thinking. Barbara Wasson from Bergen’s Centre for the Science of Learning and Technology noted that some researchers are already using AI to edit and improve their writing, stressing the importance of good communication with chatbots and strong critical awareness.
Morten Goodwin, professor at the University of Agder, emphasized ChatGPT’s short-term benefits as a writing assistant, helping draft abstracts, improve text flow, identify logical flaws, and suggest additional arguments. However, he warned that AI does not yet have the scientific rigor of human researchers and should be viewed as an intelligent but informed sparring partner. Looking ahead, he expects specialized AI tools tailored for scientists to emerge, capable of scientific writing, referencing, experiment planning, and hypothesis validation, though these remain future developments.
Finnish universities have adopted similar approaches. At the University of Helsinki, where new regulations on large language models were approved in mid-February, Professor Kai Nordlund described ChatGPT as both a change and an opportunity in university teaching. The university encourages instructors to use AI tools and prepare students for a future where AI methods will be widespread. However, use may be limited when it does not support learning. Students must declare any AI assistance used in their work, specifying the model and how it was employed. AI cannot be credited as an author since it cannot take responsibility for content, which remains the human user's responsibility. Professor Teemu Roos of Helsinki’s computer science department called these guidelines sensible, affirming that traditional rules on copying and paraphrasing also apply to AI-generated content. The guidelines see large language models as opportunities but caution that their text can be unreliable, requiring controlled use. Future refinements are expected, especially as EU-level AI regulations and ethical policies evolve.
In Iceland, a collaborative open framework on ChatGPT has been developed by a working group led by Dr. Katrín R Frímannsdóttir from the University of Iceland, with input from all Icelandic universities. The framework acknowledges AI’s potential to simplify academic work but insists its use must meet technical and ethical standards. Using AI in academic work is subject to the same rules as other sources or assistance, with misuse treated as academic misconduct. Origins must be cited according to university guidelines. Jón Atli Benediktsson, rector of the University of Iceland, emphasized the importance of universities working together to form a shared understanding.
When discussing the current limitations of AI, Dr. Kristinn R Thórisson of Reykjavik University, who has researched AI for over 30 years, explained that modern deep learning systems, including ChatGPT, rely on statistical methods and do not truly understand their generated content or the world. Students using such technology to complete homework or exams are doing “double damage”: wasting teachers’ time with mindlessly generated work and missing out on the educational process exams aim to ensure. Moreover, these AI tools often produce incorrect information, making it necessary to verify their outputs, which defeats their intended convenience. Thórisson suggested that while current AI cheating methods pose limited threats, the real challenge lies ahead when “real AI” — machines with genuine intelligence — emerges. Preparing for that future will require thoughtful adjustments across society, including in education.