Evidence-based advancement of teaching AI in K-12: an action research approach

AI is omnipresent in our daily lives. It is, therefore, crucial that students acquire necessary competencies as part of their CS education in order to be able to use and develop this technology responsibly. However, this growing need has hit the educational landscape mostly unprepared. Curricula are only gradually adapted, and there is a lack of empirical evidence on how the topic can be implemented in K-12 education. The study presented in this article uses the cyclical and participatory approach of action research to address this gap. This ensures that theories found about teaching and learning processes can be implemented directly into practice to develop AI teaching on an empirical basis. The initial cycle focuses on content-specific difficulties experienced by learners. First findings indicate that, besides general barriers such as required mathematical and programming skills, students encounter problems when applying or transferring the concepts they have studied.