The Relationship Between Artificial Intelligence Literacy and Research Competence Among Undergraduate Students in Public and Private Universities in Mogadishu, Somalia: The Mediating Role of Critical Evaluation Skills

Abstract Background: The swift penetration of generative artificial intelligence (GenAI) across tertiary education presents transformative affordances alongside epistemic vulnerabilities. While global scholarship increasingly inspects AI integration, empirical investigations contextualized within post-conflict, emerging higher education ecosystems in Sub-Saharan Africa remain remarkably scarce. Objectives: Grounded in Bandura’s Social Cognitive Theory and the Cognitive Information Processing Model, this empirical study investigates the direct relationship between Artificial Intelligence Literacy (AIL) and perceived Research Competence (RC) among undergraduate students in Mogadishu, Somalia, while evaluating the mediating mechanism exerted by Critical Evaluation Skills (CES). Furthermore, it investigates structural invariance across public and private university archetypes. Methods: A cross-sectional explanatory survey design was employed. Empirical data were collected through a validated, multi-item psychometric questionnaire administered to undergraduate students enrolled across public and private universities in Mogadishu. A total of N = 432 clean responses were retained after screening. Partial Least Squares Structural Equation Modeling (PLS-SEM) and PLS Multi-Group Analysis (PLS-MGA) were conducted using SmartPLS 4 to assess the measurement and structural parameters. Results: The empirical findings demonstrate that AI Literacy exerts a statistically significant positive direct effect on Research Competence (β = 0.284, t = 5.912, p < 0.001) and an even stronger direct impact on Critical Evaluation Skills (β = 0.542, t = 13.881, p < 0.001). In turn, Critical Evaluation Skills significantly bolster perceived Research Competence (β = 0.461, t = 9.428, p < 0.001). Mediation bootstrapping confirms that Critical Evaluation Skills partially mediate the relationship between AI Literacy and Research Competence (βindirect = 0.250, t = 7.742, 95% BCa CI [0.189, 0.315], p < 0.001), accounting for approximately 46.8% of the total effect (Variance Accounted For, VAF). Multi-group analysis reveals no statistically significant structural variance between public and private universities, substantiating the cross-institutional stability of the structural relationships. Conclusions & Implications: AI literacy alone is an insufficient condition for cultivating robust research competence; its academic utility is fully unlocked only when mediated by rigorous epistemic verification and critical scrutiny of algorithmic outputs. Higher education institutions and policymakers in Somalia must prioritize institutional AI literacy frameworks that couple prompt engineering and technical utilization with rigorous source verification, epistemic vigilance, and ethical inquiry.