Artificial Intelligence Applications in Primary Education: A Quantitatively Complemented Mixed-Meta-Method Study

Abstract

In recent years, rapidly advancing technology has reshaped our world, holding the potential to transform social and economic structures. The United Nations' Sustainable Development Goals (SDGs) provide a comprehensive roadmap that promotes not only economic growth but also social, environmental, and global sustainability. Meanwhile, artificial intelligence (AI) has emerged as a critical technology contributing to sustainable development by offering solutions to both social and economic challenges. One of the fundamental ideas is that education should always maintain a dynamic structure that supports sustainable development and fosters individuals equipped with sustainability skills. In this study, the impact of various variables related to AI applications in primary education at the elementary school level, in line with sustainable development goals, was evaluated using a mixed meta-method complemented with quantitative analyses. Within the framework of the mixed meta-method, a meta-analysis of data obtained from studies conducted between 2005 and 2025 was performed using the CMA program. The analysis determined a medium effect size of g = 0.51. To validate the meta-analysis results and enhance their content validity, a meta-thematic analysis was conducted, applying content analysis to identify themes and codes. In the final stage of this research, to further support the data obtained through the mixed meta-method, a set of evaluation form questions prepared within the Rasch measurement model framework was administered to primary school teachers. The collected data were analyzed using the FACETS program. The findings from the meta-analysis document review indicated that AI studies in primary education were most commonly applied in mathematics courses. During the meta-thematic analysis process, themes related to the impact of AI applications on learning environments, challenges encountered during implementation, and proposed solutions were identified. The Rasch measurement model process revealed that AI applications were widely used in science and mathematics curricula (FBP-4 and MP-2). Among the evaluators (raters), J2 was identified as the most lenient rater, while J11 was the strictest. When analyzing the AI-related items, the statement I can help students prepare a presentation describing their surroundings using AI tools (I17) was identified as the most challenging item, whereas I understand how to effectively use AI applications in classroom activities (I14) was found to be the easiest. The results of the analyses indicate that the obtained data are complementary and mutually supportive. The findings of this research are expected to serve as a guide for future studies and applications related to the topic, making significant contributions to the field.