Responsible integration of generative AI and LLMs in higher education: Empirical insights and a framework for teaching and learning

Abstract

Generative AI (GenAI) tools are rapidly reshaping university learning environments, yet pedagogical and ethical frameworks have not developed at the same pace as their adoption. Adopting a multidisciplinary approach, this study combines empirical evidence from a survey of 117 undergraduate Engineering and Health Sciences students with the pedagogical perspectives of university lecturers. It examines how students currently use GenAI and how these tools can be meaningfully integrated into teaching and learning. Findings show that students’ first exposure to GenAI is mainly driven by informal social media channels rather than institutional guidance, resulting in uneven and often unguided adoption. The study also identifies a “contradictory profile” in 11% of the sample: students who perceive GenAI use as a form of cheating while also expressing a strong demand for formal training on its academic integration. From the lecturers’ perspective, GenAI offers opportunities to support content creation, assessment design, and instructional workflows, provided that human-led teaching remains central. By integrating observed student behaviors with expert recommendations, this research proposes an ethical framework and practical guidelines for the responsible integration of GenAI in higher education. The study highlights the urgent need for institutional policies that address emerging socio-ethical tensions and ensure that GenAI enhances, rather than circumvents, foundational knowledge, academic integrity, and critical thinking.