A Systematic Review: The Conditions Shaping GenAI's Impact on Iranian Education

This systematic review synthesizes 15 empirical studies published in 2026 to examine how generative artificial intelligence (GenAI) is affecting Iranian education and what the current evidence supports. Following PRISMA 2020 guidelines, a search of four databases yielded 710 records, of which 15 empirical studies met the inclusion criteria and were included in a thematic synthesis. The corpus comprises twelve journal articles and three chapters in edited volumes. The synthesis identified four themes. The first is that GenAI is effective when it is paired with human mediation and less effective when it replaces it, a finding that is replicated across every study in the corpus that compared AI-only and hybrid conditions. The second is that the benefits of GenAI are unevenly distributed across linguistic domains, with stronger effects on fluency, complexity, and self-efficacy than on accuracy or global proficiency. The third is an affective asymmetry, in which learners gain in enjoyment, well-being, and self-efficacy while teachers bear new costs in the form of AI-induced anxiety and monitoring workload. The fourth is a gap between individual adoption and institutional readiness, in which teachers and students are willing to use AI while the institutional structures that would make this use sustainable have not been built. The review argues that the positive impacts reported in the literature are real but conditional, produced in designed settings that most Iranian institutions have not yet created, and that closing this gap requires hybrid models of instruction, localized tools, AI literacy training that addresses the affective dimensions of adoption, and policy frameworks suited to under-resourced higher education.