Students' Perceptions of Generative AI for University Assignment Preparation: Evidence from Higher Education in Yemen
Abstract
The incorporation of Generative Artificial Intelligence (GenAI) in higher education is revolutionizing academic preparation and support. Although global research is growing, empirical evidence regarding students’ perceptions in resource-constrained environments remains scarce. This study investigates undergraduate students’ perceptions of GenAI for university assignment preparation at the University of Science and Technology (UST), Aden, Yemen. Utilizing a quantitative cross-sectional design, data were collected from a sample of 102 students specializing in artificial intelligence. Inferential learning support revealed that students perceive GenAI as a vital academic support tool with highly significant benefits (p < 0.001). Results provided very large effect sizes for AI efficiency (M = 4.06; SD = 0.74; p < 0.001), learning support (M = 4.00; SD = 0.59; d = 1.69), assignment completion speed (d = 2.31), and co-understanding (r = 0.015; p = 0.164). Significant concerns about the loss of critical (R²=0.099) and p (β=−0.311001) and correlations between perceived concerns and utility were negligible (r=0.015, p=0.884). In an exploratory regression analysis (R² = 0.099), ethical concerns emerged as the primary significant predictor of future adoption (β = −0.311, p = 0.008), while other measures showed positive but non-significant influences in this model. The results suggest a potential "ethics–adoption paradox" with most students acknowledging the potential ethical and pedagogical risks but nevertheless wanting to adopt efficiency innovation by relying on cognitive scaffolding for learning support (0.74; it's 0.001). "Despi" (1.69), assigned to the small, single-instance, and co-understanding (rample and cesi 0.164). The study provides unique evidence of how students in developing contexts proactively adopt GenAI. Based on these findings, we propose a preliminary strategic roadmap for institutional governance, including formal AI literacy training, faculty development, and assessment redesign to prevent the erosion of intellectual agency amidst technological progress.