ChatGPT in Education: Beyond the Boon-or-Bane Debate—A Critical Narrative Review of Opportunities, Risks, and Responsible Integration

Abstract

Generative artificial intelligence, particularly ChatGPT, has rapidly entered educational practice and has intensified debate about whether it strengthens or weakens learning. This paper critically reviews the educational roles, benefits, risks, and implementation lessons associated with ChatGPT, drawing on the literature assembled in the original manuscript. The review indicates that ChatGPT can support personalized explanation, rapid formative feedback, lesson and assessment design, language learning, idea generation, and access to learning support beyond normal classroom hours. At the same time, its educational value is constrained by inaccurate or fabricated outputs, overreliance, weakened independent reasoning, ambiguity around authorship, academic integrity concerns, privacy and equity issues, and the risk of superficial learning. The evidence, therefore, does not support a simple 'boon versus bane' dichotomy. A more useful interpretation is that outcomes depend on how the technology is positioned within pedagogy, assessment, institutional policy, and human oversight. The paper proposes the Purpose–Verify–Disclose–Demonstrate (PVDD) human-in-the-loop framework, built around four principles: purposeful use, verification, transparent attribution, and authentic assessment. Practical implications are presented for students, educators, institutions, and policymakers. The paper concludes that ChatGPT is most educationally valuable when it augments rather than substitutes for human judgment, disciplinary reasoning, creativity, and teacher–student interaction.