Artificial intelligence in education: Possibilities and challenges for pedagogical practice

Abstract

Artificial Intelligence (AI) has brought about significant changes in the educational landscape, driven primarily by the advancement of generative models and the increasing availability of tools capable of supporting teaching and learning processes. This study aimed to analyze the main possibilities and challenges associated with AI in pedagogical practice through a structured narrative literature review. Searches were conducted in the Web of Science Core Collection, Scopus, ERIC, SciELO, and Google Scholar, prioritizing publications from 2019 to 2026, while earlier foundational sources were retained when methodologically or conceptually relevant. The final interpretive corpus comprised 31 core sources, including peer-reviewed empirical studies, systematic and scoping reviews, meta-analyses, and institutional guidance. Evidence was synthesized into four themes: evolution and educational applications of AI, pedagogical potential, ethical and institutional challenges, and the changing role of teachers. The literature identifies opportunities for personalized learning, pedagogical planning, accessibility, formative assessment, and teaching-material development, but also recurring concerns involving information reliability, academic integrity, privacy, algorithmic bias, digital inequality, and teacher preparedness. Recent evidence indicates that positive outcomes are heterogeneous and depend on pedagogical scaffolding, human verification, institutional governance, and AI literacy. AI should therefore be understood as a supportive educational technology rather than a substitute for teacher mediation. Its integration requires critical, ethical, and pedagogically grounded use that preserves student autonomy, assessment validity, and educational equity.