Abstract
The rapid proliferation of Generative Artificial Intelligence (GenAI) platforms such as ChatGPT, Claude, and Gemini has fundamentally transformed the higher education landscape, presenting concurrent opportunities for personalized learning and challenges to academic integrity. This study investigates the impact of GenAI integration on student cognitive engagement, self-directed learning, and academic honesty within Pakistani higher education institutions (HEIs). Utilizing a concurrent embedded mixed-methods research design, quantitative data were gathered from a stratified sample of undergraduate and postgraduate students (N = 482) across public and private universities in Pakistan, complemented by qualitative semi-structured interviews with students and academic faculty (N = 24). Structural Equation Modeling (PLS-SEM) revealed that GenAI tool utilization significantly predicts enhanced cognitive and behavioral engagement (β = 0.42, p < 0.001), particularly as an equalizer for non-native English speakers in academic writing and conceptual scaffolding. However, thematic analysis highlighted pervasive uncertainty regarding institutional guidelines, fear of false plagiarism accusations due to unreliable detection software, and varying ethical standards among departments. The findings emphasize that punitive measures are ineffective and advocate for clear institutional frameworks, assessment redesign, and AI literacy programs tailored to the socio-technical realities of developing higher education systems.
