AI boosts student feedback and grading support at the University of Bergen

An AI platform is making teachers at the University of Bergen a lot more efficient, helping them give rapid individual student feedback and accurate grade support. Thereby enhancing learning quality and motivation, freeing up time for academic professionals to focus on teaching and research.

One of the biggest challenges for students in today's higher education is the lack of good individual feedback from professionals. Traditional assessment methods are time-intensive and often fail to scale effectively, resulting in delays and generic responses that impede student development and, in worst-case scenarios, demotivate students.

A new project at the Faculty of Medicine, University of Bergen (UiB), may have found a solution to the challenge using artificial intelligence (AI). An AI platform gives students rapid, individualized feedback on formative assignments, enhancing learning quality and motivation. In exam settings, the platform offers decision support within regulatory frameworks, contributing to consistent grading support while freeing up time for teaching and research.

Every year, NOKUT (the Norwegian Agency for Quality Assurance in Education) examines the quality of higher education in Norway. For example, the NOKUT's Study Barometer shows that students at the Faculty of Medicine at the University of Bergen (UiB) are consistently dissatisfied with the academic feedback and supervision, particularly regarding the frequency of academic feedback. This trend of inadequate assessments has been ongoing over time and is also evident on campuses outside of UiB. Feedback and supervision in general within the study programme are low and have remained unchanged over time.

This was the background for the project, which was initiated by Professor Olav Tenstad at the Faculty of Medicine, UiB, together with co-inventor Christian Bru. The idea was to develop an AI-powered learning platform to enhance student satisfaction and reduce the assessment workload for academic staff.

Pictured above: Developer Christian Bru at Fjordbyte, business developer Erik Røed at Fjordbyte, and professor Olav Tenstad at UiB.

AI provides feedback and grading as good as humans

Christian Bru, Project Lead and Developer at Fjordbyte, says the idea was to test whether UiB could utilize artificial intelligence to provide faster and more accurate feedback, along with precise grading for students. The goal was to conduct a real-world test at exams for approximately 200 students at the Faculty of Medicine, with 25 assignments per student, resulting in a total of 5,000 academic feedback comments from the educational personnel.

To address this issue, UiB, with the assistance of Fjordbyte, Antire, and Tietoevry, developed a straightforward AI-based application and made it accessible on UiB's website. The feedback platform was connected to several data sources in a unified database and employed OpenAI's GPT models to generate personalized student feedback.

Fjordbyte is a startup supported by UiB, VIS (the innovation hub of Western Norway), and Innovation Norway, which owns the solution and will further develop it. Fjordbyte is poised to bring this innovation to a broader academic audience. The goal is to develop an independent, scalable, and sustainable model that can be adapted to different educational institutions, thereby further enhancing the quality of education and student satisfaction through the use of artificial intelligence.

Olav Tenstad, Project Owner and Professor, Faculty of Medicine, University of Bergen, says, "Antire quickly understood the academic context we work in. They brought the technical expertise we lacked and helped translate our pedagogical needs into a working prototype that already shows great promise for both students and academic staff."

Tenstad adds, "What impressed me most was how seamlessly Antire collaborated with our team. They weren't just consultants—they actively co-developed a tool that aligns with our educational goals and has the potential to transform how we deliver feedback." 

5,000 assessments within 10 minutes 

The AI system produced approximately 5,000 individualized grading and feedback reports in under 10 minutes in the real-world pilot test. This remarkable efficiency demonstrated the potential of AI to significantly reduce the time and effort required for grading and feedback. 

Olav Tenstad says, "This collaboration with Antire has made it possible for us to give students individual, high-quality feedback within minutes. It's raised the bar for what we can expect from digital assessment tools."  

The system provides complementary feedback on each exam question and a detailed summary of the entire exam answer, highlighting weaknesses and strengths and areas for improvement for the student. The assessment is available to both students and professors.  

Christian Bru adds, "The big question is, is this system reliable? Can we trust the grades given? Is the feedback valuable? We compared the AI results with self-evaluations, peer evaluations, and teacher evaluations to answer the questions. The results showed that AI was at least as good as the feedback and evaluation from two humans." 

The AI platform also helps academic personnel automatically generate exam questions based on learning objectives, and comprehensive statistics to evaluate question quality and set grade thresholds. 

Future steps and taking AI further 

Buoyed by the pilot project's success, the University of Bergen is exploring opportunities to expand the use of AI-driven feedback across various departments and disciplines.  

The introduction of an agent-based architecture as an alternative to the direct use of large language models, exploring the addition of interactive learning features in future iterations. 

Olav Tenstad concludes, "Thanks to Antire, we are now well underway in developing a system that saves time, improves quality, and supports our vision for education that is both research-based and student-oriented."

This project demonstrates how AI can streamline academic workflows, enhance the quality of feedback, and improve the overall educational experience for both students and professors.

This initiative is a testament to UiB's dedication to embracing technological advancements in education. By integrating AI into the feedback process, the university addresses existing challenges and sets a precedent for other institutions seeking to innovate and improve their educational practices.