According to NTNU, it has always been permissible to use Chat GPT in exams, as long as you state it as the source. Now they have changed the exam regulations.
In February last year, NTNU found it necessary to update the regulations on cheating in exams, in order to take account of the new artificial intelligence (AI) technology that quickly came onto the field.
The aim of the rule change was for it to be clear that answers generated with the help of AI, such as Chat GPT or similar, and submitted as a separate answer, would be cheating. Following a rector’s decision, the following sentence was added to the cover template for home exams:
“It is not permitted to generate an answer using artificial intelligence (Chat GPT or similar) and deliver it in whole or in part as a separate answer”
Facts
This is what NTNU’s regulations say about AI tools on exams now
“Plagiarism. This means that you use text or content that someone else has created and pretend that you have written or created it yourself. It is, for example, plagiarism to use material that someone else has created without using source references. It is also cheating to copy from online sources, other people’s assignments, textbooks, articles without clearly marking the quotation and where the quotations or content is taken from. The same applies if you use an artificial intelligence tool as a source (ChatGPT or similar) without reference to the tool as a source.”
The same text was also inserted on NTNU’s info page about cheating in exams, on Innsida.
This was interpreted by several as meaning that all use of AI was defined as cheating at NTNU: Newspaper Khrono wrote that “NTNU prohibits the use of Chat GPT in exams”. The address wrote “NTNU has banned the use of Chat GPT on exams”. VG wrote that “Norway’s largest university is clear that the use of ChatGPT or other artificial intelligence is considered cheating or plagiarism”, indeed a few weeks before the rector’s decision was sent out.
NTNU now specifies that using KI and Chat GPT has never been prohibited.
Clarification was needed
– It is not right that we went in to categorize all use of Chat GPT as cheating. The rector’s decision was that creating text using artificial intelligence and passing it off as one’s own was to be considered cheating. Use of KI must be stated in the same way as other sources, says Pro-Rector for Education at NTNU, Marit Reitan.
A working group at the Faculty of Information Technology and Electrical Engineering determined in October 2023 that there was uncertainty in the interpretation of the vice-chancellor’s decision.
– It would have been very unfortunate if the Rector’s decision on the use of generative AI is interpreted so that the use of the new tools is prohibited, the working group wrote in the report.
After discussions in the Education Committee in February this year, the wording from the rector’s decision has now been removed, and the examination regulations have been updated with a new wording that emphasizes that you must cite sources if you use AI tools (see fact box in the margin).
– There was a need for clarification, says Marit Reitan.
– KI gave us challenges almost overnight
Reitan says that when the Education Committee in the first round discussed questions related to the use of Chat GPT and cheating on exams, there were different perceptions of challenges related to the use of artificial intelligence.
Replaces the learning assistant with ChatGPT
– It wasn’t so strange either, as KI almost overnight gave us new and big challenges in terms of exams and assessment, says Reitan.
She believes that the focus in professional circles has become too much on cheating, and too little on the opportunities that AI provides and the skills development needed to support this.
– The Department for Education immediately initiated measures to develop the competence of our subject teachers. We organize workshops and seminars, and we are now working on further mapping the needs for general skills development, tools and learning support, says Reitan.
The university library has also included the use of AI in its courses on source citation and referencing, and Reitan states that a course on the use of AI in learning for students will be organized in the autumn.
Use of KI does not necessarily result in a passed exam
Despite the fact that it has now been clarified that the use of AI is permitted in the exam, Reitan cannot say anything about in which situations it is okay to use AI tools in the exam.
– The professional communities must answer for that, there are different issues within different professional areas.
This is how students use KI during exam time
– Can a student be caught cheating if large parts of the text are AI-generated, even if one shows that AI has been used?
– If AI tools are stated as the source where they are used and in such a way that it is clear what is AI-generated, there will be no cheating. Whether it leads to passing the exam is another matter, Reitan replies.
Thought it wasn’t allowed
When UA goes around the campus to ask the students, few have realized that Chat GPT has always been allowed.
Students Olav Førland and Sigurd Kampevold are currently writing their master’s thesis at Indøk with a specialization in applied economics and optimization.
– Do you know what the university’s regulations are regarding the use of KI?
– I do not know. Grammarly and that kind of thing is probably allowed, but Chat GPT should not be used as far as I know, Førland guesses.
Frøland refers to Kampevold as the most informed of the two, and he says that he knows that they must now declare their use of AI. The technical faculties have drawn up their own forms where students can answer how they have used AI in their exams or bachelor’s and master’s theses, but this has not yet been established centrally at NTNU.
Furthermore, the students say that they like to use AI for discussion or concrete coding that cannot be seen again directly in the assignments they deliver.
– What do you think about the updated guidelines?
– I am happy to hear that the regulations have been changed a little. I think it is good that it should be possible to use the tool as long as you show how you have done it, says Kampevold.
Surprised that it is legal as long as you refer
Hannah Espelin-Rolfsen and Edgar Aksel Tandberg are both in their fourth year at Industrial Chemistry and Biotechnology. They say that they did not know anything about the regulations for the use of AI, but Tandberg has heard about the declaration that is now being made.
– It varies how I myself use AI, some lecturers say outright that “here you just have to use AI”, others seem more against it. I usually just follow what the lecturers say, says Espelin-Rolfsen.
Both of the two students were surprised that all use is now legal as long as you refer to how you have used a given AI tool, but agree that it is perfectly reasonable that one must show the use of AI in their work on the same level as other sources.
