Generative AI in Education: A Systematic Literature Review on AI Literacy, Human–AI Collaboration, and Pedagogical Transformation

This review synthesises empirical and conceptual evidence on how generative AI reshapes education across three intersecting axes: AI literacy measurement, human–AI collaboration, and pedagogical transformation via assessment redesign and governance. We map instruments and frameworks used in K‑12 and higher education (2019–2026), catalogue measurement types and psychometric metadata, and compare teacher–AI partnership models and assessment redesign approaches. The evidence indicates AI literacy is consistently conceptualised as multidimensional—knowledge, application, evaluation, ethics—but studies tend to rely on self‑report scales, with few validated performance‑based tests linking literacy to demonstrable learning outcomes. Frameworks that position AI as a co‑intelligent partner or explainable tutor (DOT, PEARL, AI‑CIC) show stronger learning affordances when instructor scaffolding is present. Assessment redesign proposals that foreground process evidence and AI interaction traces (AFCL, EUIA) offer practical pathways to preserve validity and integrity, but empirical trials remain limited. Institutional policy responses are often decentralised and inconsistent, and equity and access constraints moderate reported benefits. We therefore recommend prioritising validated performance measures, experimental evaluation of assessment‑redesign models, coordinated faculty development, and coherent governance coupled with equity safeguards.