Norway

With the recent rise of AI technology, the future of education seemed uncertain: Anything seemed possible: from AI taking over the job of the teacher to a return to handwritten exams to prevent cheating. The new possibilities, but also the emerging issues, raised questions for the organisers of the history competition in the EUSTORY network: How to assess whether students had used AI in their research? Would the judging criteria need to be changed? And, looming over all of this, would students let ChatGPT write their essays?

Those were also the issues Karsten Korbøl, organiser of the Norwegian History Competition “My family in history”, wrapped his head around. As a teacher for History and Civic Education and teacher trainer at the University of Oslo he decided to test tools like ChatGPT himself to see what AI can or cannot do. He compiled his findings in an article titled “Is it possible to outsmart a chatbot?” His answer? Yes – for now. He is convinced that there are certain tasks, such as evaluating historical sources, discussing the significance and relevance of events or creative challenges are the ones that AI cannot successfully solve.

There’s a fourth task, where ChatGPT proves to be inferior: Personal approaches. Writing microhistory, such as family history, is beyond its skill, as it lacks the sources and interpretation of the material. This conviction is also mirrored in the competition’s approach on AI.

When the latest history competition started in summer 2023, the organisers decided against including AI in the competition rules or even from prohibiting students to use tools like ChatGPT. Instead, Korbøl firmly believes that this is not necessary. He relies on trust in tutors, jury members and most importantly in students themselves.

As the Norwegian History Competition only accepts entries on family or local history that use a bottom-up perspective, it severely limits how AI could be used by students. Even more importantly, why would students let ChatGPT write their family history, when the tool doesn’t know anything about the persons dealt with? In the end, the competition is not about writing the longest or most elaborate essay on general history, but about research on participants families.

But this is not the only reason for Korbøl's decision. At a recent EUSTORY organisers meeting he presented the Norwegian approach that works because of the specific set-up of the competition.

First of all, there’s only little prize money, which gives little incentive to cheat, as the promise of financial gain is low. However, stakes are high regarding students’ standings with their teachers. As the research is done usually in history classes, history teachers are also the first ones to grade the competition essays. This means that students would have to trick not only an anonymous jury, but also their own teachers, who know their writing style and personal backgrounds. If a teacher caught a student cheating on his or her research, that might severely damage the student’s reputation in class. Korbøl also hopes that teachers will inform the jury if they notice anything suspicious in students’ writings.

There are also more philosophical reasons, why Korbøl decided against an AI policy. Wouldn’t the sole mention of allowed and disallowed usage of AI encourage students to use tools like ChatGPT? And, more importantly, what is the difference between referring to AI for help and asking one’s parents for advice? In the end, AI might even provide equal opportunities to kids from different social backgrounds.

What’s important for Korbøl is the focus on historical sources. Because at the end of the day, Chat GPT is a tool and not a source and should be treated as such.

When the jury met to assess this year’s entries, Korbøl’s convictions proved to be true.

Instead of essays written with Chat GPT, the jury found well researched and narrated stories. "The level of entries in the secondary school category was very high, perhaps the highest for many years," said members of the jury. There definitely has been a positive development in the student’s evaluation and discussion of historical sources, concluded the jury. All winning entries united one thing: a clear research question and good use of historical sources. Moreover, the personal commitment that could be felt in all winning works showed that the told story mattered to the participants.

This year, many entries told the stories of women, reflecting on gender roles, traditions and different life paths. The very personal entries managed to open multiple perspectives by discussing and interpreting various sources – something AI can not do - yet.