Beyond AI Adoption: Teacher-Role Transformation and Pedagogical Adaptation in African Universities

Purpose. This study examines how the use of generative artificial intelligence (AI) is associated with university teachers’ perceived role transformation and their intention to adapt their teaching practice in African higher education. Approach. Survey data from 240 university teachers across 55 countries and all five African Union regions were analysed using partial least squares structural equation modelling (PLS-SEM) in SmartPLS 4. Measurement quality, structural paths, indirect effects and out-of-sample prediction were assessed. Post hoc analyses based on exported PLS latent scores tested a competing direct path, a fuller rival model, alternative specifications of the role construct, control variables and differences between subgroups defined by prior AI training. Results. AI competence and institutional readiness were positively associated with pedagogical AI use. Use was associated with role transformation, which remained the strongest predictor of adaptation intention. In the direct-path robustness model, both role transformation (β = .412, p < .001) and use (β = .191, p = .002) predicted intention. In the fuller rival model, the direct effect of use became non-significant, while the indirect path through role transformation remained positive (β = .135, 95% CI [.071, .210]). Removing the two potentially overlapping role items did not materially alter the central result. Predictive relevance was positive, although PLS-SEM did not outperform the linear benchmark for adaptation intention. Contribution. The findings position teacher-role transformation as a robust interpretive mechanism between AI-enabled practice and intended pedagogical adaptation. Because recruitment was non-probabilistic and the design cross-sectional, the results describe associations within this sample rather than national prevalence or causal effects.