Transformative Roles of Generative Artificial Intelligence as a Co-Instructor: A PRISMA-Guided Systematic Literature Review for Personalized and Sustainable Higher Education

The rapid advancement of Generative Artificial Intelligence (GenAI), particularly Large Language Models (LLMs) such as ChatGPT, has significantly transformed instructional practices in higher education. Beyond serving as instructional support tools, GenAI systems are increasingly conceptualized as collaborative co-instructors capable of assisting educators in curriculum planning, personalized instruction, formative assessment, and learning analytics. Despite growing scholarly interest, existing reviews often examine these applications independently and provide limited integration of pedagogical, ethical, governance, and sustainability perspectives. This study addresses this gap through a PRISMA-guided systematic literature review of 20 empirical and evidence-based scholarly publications published between 2023 and 2025. Articles were retrieved from Scopus, Web of Science, ERIC, and Google Scholar using predefined search protocols and were analyzed through directed qualitative content analysis. The synthesis identified three dominant instructional functions of GenAI: co-planning, co-instruction, and co-assessment, each demonstrating measurable improvements in instructional efficiency, learner engagement, personalized learning, and formative feedback. Simultaneously, the review highlights significant challenges related to academic integrity, algorithmic bias, data privacy, teacher autonomy, and student overreliance on AI-generated content. Comparative analysis across studies indicates that effective implementation consistently depends on sustained human oversight through a Teacher-in-the-Loop (TiTL) framework, which positions educators as pedagogical decision-makers while leveraging AI to augment instructional effectiveness. The review further demonstrates that responsible GenAI integration contributes to sustainable higher education by improving resource efficiency, reducing faculty workload, expanding equitable access to learning, and supporting resilient digital education ecosystems. The study contributes a synthesized conceptual perspective that integrates pedagogical, ethical, governance, and sustainability dimensions into a unified framework for responsible AI-augmented higher education. The findings provide evidence-based guidance for educators, institutional leaders, and policymakers seeking to implement GenAI responsibly while preserving academic quality, educational equity, and human-centered teaching.