Estonia identifies 12 risks of implementing AI in schools — ERR News
A working group at the Centre for Ethics of the University of Tartu in Estonia has published an analysis of the risks and value choices associated with the AI Leap (TI-Hüpe) education programme and the creation of an artificial intelligence-based learning application. The document identifies 12 risk scenarios and provides recommendations for the programme, the state, schools, research institutions and technology providers, ERR News reports.
The risks were divided into pedagogical, societal, and infrastructure and governance risks. They concern not only the system’s technical operation and cybersecurity, but also learning objectives, the reliability of assessment, the teacher’s role, data protection, and the preservation of the Estonian language and culture.
AI must deepen learning
At the beginning of 2024, 90% of Estonian upper-secondary students used artificial intelligence in their studies, mainly free commercial applications, to complete school assignments faster. The authors of the analysis warn that a ready-made high-quality result does not always mean that a student has mastered the material, can explain the logic of the answer, or apply the knowledge in another situation.
As part of AI Leap, OpenAI and Estonian researchers are developing a Socratic learning application for schools. Unlike a regular chatbot, it is intended to help students reach conclusions independently through questions, prompts and feedback. At the same time, the effect will depend on how the tool is integrated into the learning process, what tasks are selected for it, and how clearly teachers can see students’ work process.
More current news is available on the UA.News Telegram channel Telegram.
The researchers also pointed to the need to change assessment: bans and monitoring alone are insufficient to ensure academic integrity. Tasks should reveal students’ thought processes, and schoolchildren should be taught to honestly disclose their use of AI, verify its answers and take responsibility for them.
The role of the teacher and the Estonian language
The report stresses that artificial intelligence must not turn into a parallel teacher or moral authority. It does not know the student as an individual and is not responsible for their development, so the teacher must retain a leading role in the learning process. In autumn 2024, 47% of teachers in a study said they had not used AI tools in their work.
The analysis separately raises issues of values and language policy. The responses of large language models depend on training data, developers’ decisions, safety filters and commercial objectives. International models may also insufficiently account for Estonia’s curriculum, terminology, cultural context and high-quality materials in Estonian. The authors propose continuously assessing the application’s language quality, factual accuracy, cultural appropriateness and alignment with values, with the participation of universities, teachers and relevant specialists.
