The integration of Artificial Intelligence (AI) into elementary education has fundamentally altered student learning frameworks. While offering efficiency, easy access to instant solutions triggers a critical dependency that threatens children’s natural cognitive structure. This study aims to identify, analyze, and evaluate the specific impacts of AI dependency on the cognitive learning outcomes of elementary school students. Operating a Systematic Literature Review (SLR) method strictly adhering to the PRISMA 2020 protocol, metadata tracking was comprehensively executed using the Google Scholar database for articles published between 2021 and 2025. Following rigorous inclusion and exclusion screenings, 14 relevant scientific articles were ultimately selected for qualitative content analysis. The synthesis highlights a bifurcated impact of AI implementation. When systematically directed, AI significantly operates as a positive learning assistant by clarifying abstract concepts through visualization and tailored instructional personalization. Conversely, unregulated usage induces a severe dependency pattern that distinctly degrades critical thinking, analytical reasoning, and long-term memory retention. This decline heavily restricts opportunities for autonomous problem-solving and hampers Higher Order Thinking Skills (HOTS). Conclusively, AI must be strictly stationed as a learning assistant rather than a surrogate for human cognitive processing. Maximizing AI’s pedagogical potential requires a robust synergy of instructional supervision by teachers and parents, alongside cultivating foundational digital literacy from an early age.