Abstract
Geometry education in South Africa's senior phase (Grades 7–9) is characterised by persistent learner underperformance, abstract instructional approaches, and a gap between the transformative potential of emerging technologies and their realisation in classroom practice. Artificial Intelligence (AI) tools have been identified globally as capable of addressing these challenges. However, within the South African context, the perceptions and readiness of educators tasked with implementing these tools remain substantially underexplored. This study investigated senior-phase mathematics educators' perceptions of AI tool integration in geometry teaching. Informed by a constructivist paradigm and employing a qualitative case study design, the research engaged nine purposively selected educators from eight urban schools in the Free State and Gauteng provinces through semi-structured interviews. Data were analysed using the Noticing–Collecting–Thinking (N-C-T) model and ATLAS.ti software, generating 1 474 coded quotation segments across five themes. Educators expressed cautious optimism, acknowledging benefits in visualisation, time efficiency, personalised differentiation, and real-world contextualisation, while navigating infrastructural unreliability, device scarcity, and assessment misalignment. AI use remains confined to lesson preparation, with GeoGebra and ChatGPT most frequently cited. Educators identified cognitive surrender, whereby learners defer reasoning to AI outputs, as a threat to productive struggle underpinning van Hiele level progression. One educator with a physical disability reframed AI integration as a matter of professional inclusion and equity, absent from existing South African literature. A lack of subject-specific professional development was identified as the critical mediating variable between positive perceptions and classroom practice. The study concludes that AI integration in geometry represents a pedagogical and institutional challenge rather than a technological one, and recommends the urgent development of Artificial Intelligence Technological Pedagogical Content Knowledge (AI-TPACK) informed professional development programmes.