This narrative review critically examines the integration of generative artificial intelligence (AI) in higher education, addressing a knowledge gap in understanding how AI-mediated learning aligns with sociocultural perspectives of cognitive development. Despite widespread adoption of tools such as ChatGPT, existing literature largely focuses on technological capabilities or individual cognitive outcomes, overlooking the interplay between human intelligence, pedagogical practices, and collaborative knowledge construction. Anchored in Vygotsky’s Sociocultural Theory, the review explores AI as a mediational tool shaping cognitive reconfiguration, hybrid intelligence, pedagogical transformation, assessment innovation, and ethical repositioning in diverse higher education contexts. A systematic search across Scopus, Web of Science, ERIC, and Google Scholar, covering 2020–2025, employed structured keywords related to AI, higher education, and cognitive development. Selected studies underwent thematic synthesis, examining cognitive, pedagogical, assessment, ethical, and epistemological dimensions, with rigour ensured through cross-validation, triangulation, and reflexive consideration of researcher positionality. Findings indicate that AI reorients human intelligence from knowledge retention to orchestration, enhances metacognitive regulation, and supports hybrid learning, while challenges include cognitive offloading, inequitable access, and unstructured reliance that may undermine deep engagement. The review concludes that AI strengthens intellectual development when embedded in scaffolded, ethically guided, and collaborative environments. Policy implications emphasise equitable access, AI literacy, and process-oriented assessment, whereas pedagogical practice should prioritise cognitive coaching and reflective engagement. The study contributes to the body of knowledge by synthesising theoretical and empirical insights to guide AI-mediated higher education that preserves intellectual agency, fosters responsible AI adoption, and advances context-sensitive learning strategies.
Tom Ongesa Nyamboga (Fri,) studied this question.