<![CDATA[The rapid integration of generative artificial intelligence (GenAI) in initial teacher education has fundamentally disrupted conventional lesson-planning paradigms, yet its specific application within the distinct developmental boundaries of primary school teacher preparation remains highly fragmented. This systematic evidence synthesis addresses this gap by analyzing the global empirical literature published between January 2022 and early 2026 to evaluate how GenAI shapes pre-service elementary teachers' pedagogical design, technological agency, and critical evaluative literacy. Adhering strictly to the PRISMA 2020 guidelines, systematic database searches were executed across Scopus, Web of Science, ERIC, IEEE Xplore, and Google Scholar to compile a highly refined final corpus of exactly twenty-five peer-reviewed empirical studies, with methodological quality appraised using the Mixed Methods Appraisal Tool (MMAT). Qualitative and quantitative findings were synthesized using a rigorous six-phase thematic synthesis framework to map candidate competencies against theoretical frameworks of teacher agency and cognitive load. The findings demonstrate a major pedagogical transition wherein candidate teachers leverage large language models to construct customized Concrete-Pictorial-Abstract (CPA) scaffolding suitable for concrete-operational learners aged 5–12. Furthermore, prompt-assisted software authoring reconfigures candidate agency, enabling pre-service educators to write lightweight canvas code for interactive digital manipulatives and reclaim pedagogical sovereignty from proprietary commercial educational ecosystems. However, this operational velocity is governed by a critical evaluative vigilance gap and automation bias, wherein candidates' high self-efficacy frequently leads to the uncritical classroom enactment of subtle, mathematically or scientifically invalid domain hallucinations. To resolve these theoretical and ethical tensions, initial teacher education curricula must undergo systematic, multi-year reforms that shift from technical tool tutorials toward structured process-oriented auditing, epistemic verification rubrics, and child data privacy protocols during clinical placements.]]>
