Artificial intelligence (AI) literacy is a global strategic objective in education. However, little is known about how AI should be taught. In this paper, 46 studies in academic conferences and journals are reviewed to investigate pedagogical strategies, learning tools, assessment methods in AI literacy education in K-12 contexts, and students’ learning outcomes. The investigation reveals that the promotion of AI literacy education has seen significant progress in the past two decades. This highlights that intelligent agents, including Google’s Teachable Machine, Learning ML, and Machine Learning for Kids, are age-appropriate tools for AI literacy education in K-12 contexts. Kindergarten students can benefit from learning tools such as PopBots, while software devices, such as Scratch and Python, which help to develop the computational thinking of AI algorithms, can be introduced to both primary and secondary schools. The research shows that project-based, human–computer collaborative learning and play- and game-based approaches, with constructivist methodologies, have been applied frequently in AI literacy education. Cognitive, affective, and behavioral learning outcomes, course satisfaction and soft skills acquisition have been reported. The paper informs educators of appropriate learning tools, pedagogical strategies, assessment methodologies in AI literacy education, and students’ learning outcomes. Research implications and future research directions within the K-12 context are also discussed.

Artificial intelligence (AI) literacy education for young students is gaining traction among researchers and educators. Researchers are developing courses and attempting to teach AI literacy to younger students, using age‐appropriate AI educational learning tools. Although teachers play a crucial role in AI literacy education, their perceptions and attitudes have received little attention. This study explores the perceptions of 60 teachers regarding the use of AI educational learning tools, and examines the factors influencing their attitudes in relation to implementing AI literacy education. The technological acceptance model and the technological, pedagogical, and content knowledge (CK) (TPACK) framework inform the research design, and a mixed method, combining the statistical package for Social Science and thematic analysis, is employed for data analysis. The study reveals that teachers have positive perceptions regarding the usefulness and ease of use of AI educational learning tools in their AI literacy teaching. This paper also reveals that teachers embrace an arts‐based approach to teaching AI literacy. The qualitative data reveal that teachers face challenges such as insufficient CK and experience with AI; and knowledge of TPACK. The five factors affecting their acceptance of AI educational learning tools are: (a) teachers' perceptions of their AI CK and experience in teaching AI literacy (technological content knowledge); (b) technical challenges and stakeholder acceptance; (c) the attributes of AI educational learning tools; (d) school infrastructure and budget constraints; and (e) potential for distraction and negative emotional responses. This study offers insights for policymakers regarding professional development initiatives and technical support mechanisms, thereby facilitating more effective AI literacy implementation.

Artificial intelligence (AI) has become an important part of 21st-century life and this requires that students understand how to adapt to this change. Despite calls to extend AI literacy education from university to young students, there remains a lack of evidence-based research informing educators and researchers on the content of AI literacy education at the primary school level. This study aims to offer a comprehensive overview of what constitutes AI literacy education in primary schools, including its theoretical frameworks, pedagogical strategies, learning tools, assessment methods, educational outcomes, and challenges. The study presents a systematic review mapping process using Scopus and the Web of Science. Twenty-five empirical studies focusing on the primary school context were selected. The review revealed that (1) AI literacy encompasses a fluid conceptual understanding of digital literacy in terms of interacting and collaborating with AI, computational thinking, critical data literacy, and AI ethics; (2) constructionism, the constructivist theoretical framework, and the ARCS (Attention, Relevance, Confidence, and Satisfaction) model of instructional design were the dominant means of constructing the research design of the selected studies; (3) constructivist methodologies, project-based learning, programming, and human-agent interaction were commonly used by the selected studies, and AI learning tools, particularly intelligent agents, were often adopted in AI literacy teaching and learning research; (4) mixed-research methods were the most common, with surveys, interviews, and artifacts being employed to evaluate students’ learning outcomes; and (5) there were positive academic, affective, and behavioral educational outcomes, as well as course satisfaction. To our knowledge, this is the first systematic review to offer a detailed perspective on AI literacy at the primary school level, providing valuable insights for researchers, policymakers, and educators designing future educational frameworks, curricula, and assessment. Future research directions are also discussed.

Background: While the integration of robot‐based learning in early childhood education has gained increasing attention in recent years, there is still a lack of evidence regarding the impact of AI robots on young children's learning. Objectives: The study explored the effectiveness of two AI education approaches in advancing kindergarteners' computational thinking, sequencing, self‐regulation and theory of mind skills. Methods: An experiment was conducted with 90 kindergarteners (ages 5–6) randomly assigned to either a direct instruction (DI), cooperative play (CP) or control group. Results: Results show that (1) children in all three groups had significant improvements on computational thinking, sequencing and self‐regulation; (2) both early AI education approaches (CP and DI) significantly enhance young children's computational thinking, sequencing, self‐regulation and theory of mind skills; (3) the DI group had significant higher improvement than the CP group on computational thinking; (4) the CP group exhibited greater enhancements in theory of mind skills than the DI group. Conclusion: These findings jointly demonstrate that each AI educational approach has unique strengths, underscoring the significance of designing new pedagogies to expand children's skills.