AI center aims to prevent exclusion in education

"Our AI center will contribute to changes in light of the UN's Sustainable Development Goals, where inclusion and reduction of inequality in education will be central. AI tools in research and teaching can also be used in training for other professions, and we will explore how.

Together with a comprehensive consortium of researchers and partners, Professor Rune Johan Krumsvik has applied to establish a research center for artificial intelligence in the education, defense, police, and health sectors, with the goal of strengthening AI research in Norway.

A central question for him and his team is how AI will affect and change education. Can AI influence social and digital divides in society and lead to increased social mobility? Or will AI maintain or even increase inequality in society?

AI and Social Inequality 

Krumsvik points out that there is much uncharted territory within AI in education. Interaction, situational learning, and social learning in various contexts, with and without AI, will be particularly important to research to pave the way for a better foundation for digital competence for all.

Artificial Intelligence (AI) is developing at a faster pace than we can imagine, and it is crucial that the education field keeps up and adopts the right AI tools

Professor Rune Johan Krumsvik

The center will critically examine whether AI can cement and reinforce social inequality, or if AI can represent something entirely new.

"If, for example, we can lift the more vulnerable student groups with AI tools, this is something that should be researched, and we want to do that."

Sparring Partner for Learning 

Krumsvik believes that AI in 2025 is about much more than new technology.

"It is also about the fact that we have given AI the key to our entire linguistic cultural heritage. This challenges our entire education system, and research must contribute to strengthening our knowledge."

Krumsvik and colleagues have several very interesting research findings on how AI can be used in assessing students' performance and evaluating, for example, exam answers. AI can help provide feedback to students.

"The results show that AI can be used as a sparring partner for learning. In March 2023, we examined how capable and reliable the largest language model, GPT-4, is."

The researchers looked at how well it handles a demanding Norwegian-language school exam and acts as a sparring partner for writing at the doctoral level. They also trained GPT-4 to become more subject-specific, piloted it, and tested its capabilities within doctoral education.

"Our findings show that GPT-4 can be a good sparring partner for both students and doctoral candidates. It can help support assessment practices in higher education under certain conditions."

Olav Tenstad, professor at the Faculty of Medicine, University of Bergen, will lead one of the research packages at the center, and he has also conducted case studies similar to those of Krumsvik. He and Krumsvik will take this research further in the Digital Learning Communities AI Centre.

"I think it can develop into reliable and credible feedback, but such AI-generated assessments will be a supplement to the other regular assessments between student and teacher."

Groundbreaking with Large Language Models (LLM)

The center application focuses on artificial intelligence (AI) with specialization in Large Language Models (LLMs), such as GPT-4.

"These models are very versatile and user-friendly but often lack subject-specific adaptation. We will work on developing different solutions based on LLMs. In this way, we can explore AI-generated learning tools that can be tailored to the needs of students."

Krumsvik believes that AI is both an ethical minefield and a groundbreaking technology with many interesting possibilities.

"It is why it is important to research the consequences of, for the first time in history, a non-human entity using human language in a way that is so similar to how humans use it, that it becomes difficult to see whether it is a human or a machine communicating."

User Perspective is especially important

In addition to the 25 research partners, DLCAIC also has extensive collaboration with several user partners and networks.

"We need to have an expanded and deeper knowledge preparedness in society. AI research and innovation in the education field will contribute to cross-sectoral changes and strengthen society through digital competence building."

Partners UiB, UiO, University of Gothenburg, Norwegian School of Sport Sciences, Norwegian Defence University College, Norwegian Police University College, Manchester Metropolitan University, University of Bristol, Stanford University, University of California, Norwegian Institute of Public Health, Wenger-Trayner Social Learning Lab, LIVV Health, research schools NORED and GRADE, Inland Norway University of Applied Sciences, Knowledge Centre for Education, Norwegian Defence Research Establishment, Western Norway University of Applied Sciences, Volda University College, TK-Vestland, NIFU, University of Tromsø, NTNU, USN.