Human-Al Collaboration in the STEM Classroom: A Systematic Literature Review of GenAI as a Complement in Higher Education

Abstract

The rapid development in Generative Artificial Intelligence (GenAI), particularly in Large Language Models (LLMs), is reshaping higher education. This systematic literature review synthesizes findings from 94 empirical studies published in the last two years to map the current landscape of Human-AI collaboration in the STEM disciplines. The analysis reveals that LLMs are predominantly used as ‘Tutors & Assistants,’ focusing on helping improve measurable outcomes like academic performance. However, emerging adoption models cast GenAI as a ‘Collaborative Partner,’ fostering co-creation and higherorder thinking. This shift transforms the student’s role from a passive recipient to an ‘Active Agent & Critical Evaluator,’ and the educator’s role to a ‘Learning Architect.’ Key challenges, including the accuracy of AI-generated content and the need for effective pedagogical integration, remain significant. This review provides a comprehensive overview of current adoption models, domains of impact, evolving roles, and challenges, offering crucial insights for educators and researchers navigating this new educational landscape.