AI & grading

Heiko Pääbo (University of Tartu): AI as grading assistant (automated) 

Over the past eighteen months, I have used AI tools to develop self-study materials for my postgraduate political science courses. I regularly use a flipped classroom approach, where students work through assigned readings and resources independently before coming to seminars. This approach gives students flexibility to engage with materials on their own schedule, but it requires effective self-assessment tools so they can check their understanding of complex political concepts.

Creating these self-assessment tools has traditionally been very time-consuming. This led me to explore how AI could enhance this process. My main goal in integrating AI tools was to improve my students’ learning experience while making my workload more manageable. By using tools like NotebookLM and Moodle’s automated systems, I’ve created self-study materials that provide immediate, personalized feedback to students.

NotebookLM has been particularly useful for this purpose. One of its key benefits is the ability to work with study materials in multiple formats—PDF files, videos, websites, electronic books, and manually inserted texts. It synthesises content across several sources simultaneously and can generate questions and feedback directly linked to specific source materials. This traceability is essential in academic contexts, as it allows both the instructor to see exactly where the information originates. Furthermore, NotebookLM has clear privacy guidelines and does not use uploaded materials for training its language model, which is especially important when working with copyrighted or sensitive educational content.

This AI integration serves two important purposes:

First, it enhances the student learning experience. Students work through challenging political theory texts, follow online lectures about the political processes or try to synthesise various data to understand some patterns of behaviours. It is a challenging task but having assisting instruments like self-assessment tools, they can test their understanding. When they complete these tests, they receive immediate feedback explaining why their answers are correct or incorrect, along with suggestions for which resources to revisit if they need to strengthen their understanding. This process encourages reflection and gradual improvement in their analytical skills.

Second, as an instructor, I can focus more of my time on high-value teaching activities by automating routine tasks like creating basic questions and providing initial feedback. I still maintain oversight to ensure everything aligns with my course learning objectives and disciplinary standards.

Although my political science seminars typically have moderate-sized groups, using AI tools makes it possible to provide individualized feedback even for larger cohorts. This promotes independent learning while maintaining consistent quality throughout the assessment process. This capability is particularly valuable when considering the development of larger online courses or MOOCs in political science.