Mapping Generative AI Ethics in Higher Education: A Systematic Review and Multidimensional Boundary Framework

The rapid integration of generative artificial intelligence in higher education has generated a fundamental dilemma: does it function as a learning aid or as an academic shortcut? This study employs a qualitative systematic literature review guided by PRISMA procedures to map the ethical spectrum of generative AI use between cognitive augmentation and academic substitution. Thirty-eight core journal articles published between 2021 and 2026 were thematically synthesized. The findings indicate that generative AI can enhance learning when it supports metacognitive reflection, scaffolding, and self-regulated learning while preserving human evaluative control. Conversely, risks emerge when AI contribution replaces core cognitive labor, leading to authorship ambiguity, integrity violations, and superficial engagement. To reconcile these tensions, this study proposes a multidimensional ethical boundary framework structured along two dimensions: human cognitive engagement and level of AI contribution. This framework offers a conceptual basis for policy development, assessment redesign, and responsible pedagogical integration, positioning ethical AI use as a continuum grounded in sustained human intellectual accountability.