<![CDATA[The advent of generative Artificial Intelligence (AI) in the educational landscape has sparked a sharply polarized discourse: viewed either as a revolutionary instrument that accelerates literacy or as a latent threat that erodes deep critical thinking. This study aims to examine this paradox through a Systematic Literature Review (SLR) adapting the PRISMA protocol. Through a comprehensive search across the Scopus, ScienceDirect, ERIC, and Google Scholar databases spanning the years 2020–2026, a total of 25 reputable journal articles meeting the inclusion criteria were extracted and thematically analyzed. The synthesized literature reveals a distinct duality of impact. On one hand, AI serves as cognitive scaffolding that facilitates personalized learning, dismantles sociolinguistic barriers, and enhances multimodal literacy agility. Conversely, an over-reliance on AI automation fosters the phenomenon of "outsourcing of thought," superficial intensive reading habits caused by the absence of mental friction, and a severe threat to academic integrity driven by the illusion of competence. The findings assert that AI does not inherently enhance or erode literacy; rather, its impact is entirely mediated by classroom pedagogical design. Consequently, this paper advocates for a paradigm shift in educational assessment moving from product-oriented evaluation toward iterative process-oriented evaluation alongside the urgent integration of Critical AI Literacy into the national curriculum.]]>
