Beyond Tool Access: A Systematic Review of Human–GenAI Interaction Patterns, Trade-offs, and Design Principles in Higher Education

Generative AI tools have rapidly entered higher education, yet discourse focuses predominantly on broad benefits and risks rather than the specific interaction mechanisms predicting educational outcomes. This systematic review addresses how students and instructors interact with GenAI and which patterns predict success versus harm. We systematically searched Scopus, Web of Science, ERIC, Google Scholar, and ACM Digital Library for publications from January 2020 through January 2025, identifying 67 empirical studies and rigorous reviews. Evidence synthesis employed thematic analysis organized around five research questions addressing effective interaction patterns, outcome trade-offs, feedback design, practical principles, and equity implications. Meta-analytic evidence confirms large performance effects (Hedges’ g = 0.867) but reveals a critical paradox: strategic prompting, iterative dialogue, and verification behaviors distinguish productive from counterproductive use, yet unguided GenAI access produces “metacognitive laziness” where surface performance gains occur without knowledge transfer (g = 0.0), accompanied by increased procrastination and dependency. Effective interaction requires explicit scaffolding through structured prompts, metacognitive support, verification training, and instructor-mediated design. GenAI feedback style differentially affects motivation and learning based on task complexity and learner characteristics. Critical trade-offs emerged between efficiency and deep processing, product quality and procedural fluency, and personalized support and overreliance. Equity analysis revealed that underrepresented minorities, first-generation students, and Global South learners face compounding access and literacy barriers that GenAI risks amplifying. We conclude that interaction quality, not mere tool access, determines whether GenAI enhances or undermines learning. For higher education to realize GenAI’s potential while advancing SDG4 goals of inclusive, equitable quality education, institutions must prioritize interaction literacy, equitable access, and pedagogical scaffolding over technological deployment alone.