A systematic review on how educators teach AI in K-12 education

Abstract

Developing Artificial Intelligence (AI) education in K-12 contexts, i.e., teaching students about AI, is critical to promote students' AI literacy. However, the state-of-the-art of AI education is not clear enough. To this end, this study reviewed 45 high-quality empirical studies on K-12 AI education over the past decade from both research and instruction perspectives. Regarding the research design, this study revealed the relationship between publication year, sample size, learning stage, educational setting, research method, research focus and duration. Regarding the instruction design, this study revealed the relationship between learning stage, pedagogical strategy, learning tool, learning activity, learning content, assessment method and learning effect. Besides, this study also derived recommendations for research (i.e., time allocation, samples selection, longitudinal design, rigorous methodology and technical democracy) and instruction (i.e., group learning, authentic context, teacher involvement, triangular evidence and learning scaffolding). Overall, the main findings indicate that K-12 AI education has the potential to develop students’ AI literacy, which contains AI knowledge, AI affectivity, and AI thinking. However, deficiencies in research and instructional design still remain, including short durations, small sample sizes, non-standardized research methods, lack of long-term and cross-age AI curriculum, etc. This study also discussed several critical topics for future research and instruction.