Measuring AI Usage in Geometry Learning Among Pre-Service Mathematics Teachers: Scale Development and Institutional Comparison in Indonesia

Pre-service mathematics teachers engage with AI as both geometry learners and future teachers, yet domain-specific instruments for measuring this usage remain scarce. This study developed and validated the AI Usage in Geometry Learning Scale (AIGLS) and compared usage between two Indonesian teacher-education institutions. A comparative survey covered 178 pre-service mathematics teachers at Universitas Negeri Makassar (UNM, n = 89) and Universitas Negeri Jakarta (UNJ, n = 89). The AIGLS operationalises five dimensions: Conceptual Understanding Support, Problem-Solving Assistance, Geometric Visualisation, Pedagogical Readiness, and Critical Evaluation of AI. A ten-expert panel established content validity (CVI = 0.976). Item analysis excluded all five negatively worded items; all analyses therefore used the final 20-item scale, which showed good full-scale (Cronbach’s α = 0.873) and subscale (α = 0.591–0.810) reliability. ChatGPT was the dominant tool (96.6%), and UNM reported significantly higher usage intensity than UNJ (χ2(4) = 27.917, p < 0.001). Mann–Whitney U tests with Bonferroni correction confirmed UNM scored higher overall (U = 4749.5, p = 0.022, r_rb = −0.20), with Geometric Visualisation showing the largest institutional gap (r_rb = −0.38, medium effect). The validated AIGLS supports measuring AI usage in geometry teacher preparation and informs AI-literacy policy.