Generative Artificial Intelligence in Academic Writing: A Critical Review of Scholarly Perspectives, Pedagogical Implications, and Emerging Governance Frameworks

Abstract

The emergence of generative artificial intelligence (GenAI) tools most prominently large language models (LLMs) such as ChatGPT, Gemini, and Claude has catalysed one of the most consequential shifts in the landscape of academic writing in recent decades. This paper presents a critical, thematically structured review of scholarly perspectives on the use of GenAI in academic writing, synthesising peer-reviewed literature, institutional policy documents, and empirical data published primarily between 2022 and 2025. Drawing on more than 25 primary sources, the review identifies and interrogates six interrelated thematic domains: (1) the scope and nature of GenAI adoption in higher education, (2) perceived benefits including writing quality improvement and productivity gains, (3) epistemic risks encompassing hallucination, algorithmic bias, and cognitive dependency, (4) challenges to academic integrity and the limitations of detection technologies, (5) questions of authorship, transparency, and scholarly ethics, and (6) equity, access, and the emerging global digital divide. A novel synthesis is offered by situating these themes within a governance trajectory analysis that maps the evolution of scholarly discourse from early containment-oriented responses (2023) toward pedagogical redesign and AI literacy frameworks (2024–2025). The paper concludes by proposing a conceptual framework the Responsible AI Academic Writing (RAAW) Model that integrates transparency, critical literacy, institutional accountability, and equitable access as co-equal pillars for navigating GenAI's role in scholarly practice. Implications for institution policy, curriculum design, journal editorial standards, and future research are discussed.