Artificial Intelligence in Education: A Systematic Review of Pedagogical Impacts, Ethical Challenges, and Implications for Teacher Competencies and Educational Policy

Abstract

Background: Artificial intelligence (AI) is rapidly transforming educational settings worldwide, yet comprehensive reviews that simultaneously examine its pedagogical, ethical, and policy dimensions remain scarce. This systematic review addresses this gap by synthesizing evidence on how AI technologies are being integrated across educational levels and what implications emerge for student learning, ethical governance, teacher preparation, and institutional policy.

Methods: Following PRISMA 2020 guidelines, we conducted a systematic search across the Elicit academic corpus (encompassing Semantic Scholar and OpenAlex, over 138 million papers) using three semantically complementary queries targeting pedagogical impacts, ethical challenges, and teacher competencies. From 210 initially retrieved records, 161 passed abstract-level screening against three predefined criteria (educational focus, publication quality, and thematic relevance). Eighty studies published between 2020 and 2025 underwent structured data extraction across eight dimensions, and 35 were selected for the final synthesis based on citation verification and thematic contribution.

Results: Adaptive learning systems (52.5% of studies), intelligent tutoring systems (35.0%), and learning analytics (31.3%) constitute the dominant AI technologies examined. Forty-five studies (56.3%) report positive impacts on student outcomes, while 14 (17.5%) document mixed results contingent upon implementation context. Data privacy (56.3% of studies), digital equity (53.8%), and algorithmic bias (48.8%) emerge as the three most frequently cited ethical concerns. Teacher digital competency requirements appear in 68.8% of studies, yet only 40% of educators actively use AI tools, primarily due to insufficient training (60%), funding limitations (50%), and ethical concerns (40%). Policy frameworks remain fragmented, with 35 studies (43.8%) identifying regulatory gaps and 28 (35.0%) proposing new governance frameworks.

Conclusions: AI holds substantial transformative potential for education, but its effectiveness is contingent upon adequate teacher preparation, robust ethical governance, equitable infrastructure, and coherent policy frameworks. Without simultaneous investment across these dimensions, AI risks exacerbating rather than reducing educational inequalities. This review provides a consolidated evidence base and conceptual framework to guide policymakers, institutional leaders, and researchers toward responsible AI integration in education.