Researchers from Tallinn University are investigating the potential for artificial intelligence to evaluate the mother tongue final exams in Estonia. This initiative stems from a study conducted by the education and youth agency, Harno, which revealed that AI language models yield grading results comparable to those provided by human examiners. The study was spearheaded by Merilin Aruvee, a lecturer in mother tongue education and applied linguistics at Tallinn University, who has been involved in testing 9th-grade e-exams and developed new assessment criteria for writing tasks.

Aruvee and her team developed specific assessment criteria for the 9th-grade final exam writing task, prompting the question of whether AI could be integrated into the grading process. They meticulously detailed the criteria for evaluating exam papers and researched assessment models from around the world. Experienced language teachers from Harno tested these criteria to provide feedback on the AI model, which led to refinements based on educators' recommendations. The combination of AI's analytics capabilities and the teachers' expertise aims to enhance the overall assessment process of the mother tongue exams.

This initiative signifies a shift toward modernizing educational assessment in Estonia, potentially providing a more efficient and objective approach to grading. By incorporating AI into the evaluation process, the aim is to alleviate the workload for educators while ensuring that students receive fair and accurate assessments. This project may set a precedent for how technology can transform educational methodologies and assessment frameworks within the country.