Pre-service teachers’ interactions with generative AI for creating culturally grounded visual teaching resources

Abstract

Generative artificial intelligence (GenAI) offers new possibilities for developing instructional resources, but evidence remains limited on how technological confidence, cultural responsiveness and system experience relate to the quality and instructional use of AI-generated visuals in teacher education in sub-Saharan Africa. This study examined these relationships among 201 final-year pre-service teachers at Bagabaga College of Education, Ghana, after they had gained practical experience with the Culturally Responsive Visual Aid Generator (CReVAG) during teaching practicum. A post-exposure cross-sectional explanatory survey assessed Self-Efficacy, Culturally Responsive Teaching Self-Efficacy, Perceived System Usability, AI Tool Acceptance, Visual Resource Quality and Lesson Delivery Effectiveness. Partial Least Squares Structural Equation Modelling (PLS-SEM) was complemented with Necessary Condition Analysis (NCA) and robust subgroup comparisons. The model explained 39.9% of the variance in AI Tool Acceptance, 39.3% in Visual Resource Quality and 61.6% in Lesson Delivery Effectiveness, with an SRMR of 0.077. The largest structural associations were between System Usability and AI Tool Acceptance (β = 0.330, p = .004, 95% CI [0.118, 0.564]) and between Visual Resource Quality and Lesson Delivery Effectiveness (β = 0.309, p < .001, 95% CI [0.151, 0.478]). NCA identified Visual Resource Quality as the strongest statistical necessity condition for high Lesson Delivery Effectiveness (d = 0.190, p = .002), followed by AI Tool Acceptance (d = 0.158, p = .0015). Programme differences were most defensible for System Usability and Lesson Delivery Effectiveness, while gender differences were generally small, although males reported higher AI Tool Acceptance (d = 0.40, p = .006). The findings suggest that meaningful GenAI-supported instructional design involves more than favourable technology appraisal; it also requires usable systems, culturally informed teacher judgement and careful evaluation of generated visual resources. Because the constructs were measured once after exposure to CReVAG, the findings represent explanatory associations rather than causal effects.