Utilization of Artificial Intelligence to support administrative decision-making in special education institutions in Saudi Arabia: perceptions of principals, supervisors, and teachers

Artificial Intelligence (AI) is increasingly used in educational administration to automate routine tasks, support data-informed decision-making, and improve institutional efficiency. This study examined the perceptions of principals, supervisors, and teachers regarding the utilization of AI to support administrative decision-making in special education institutions and investigated whether these perceptions differed according to professional role, years of experience, and AI-related training. The sample comprised 173 participants working in public schools providing special education services in Riyadh, Saudi Arabia. A descriptive cross-sectional design was employed, and participants were recruited using non-probability convenience sampling. Data were collected using a 23-item questionnaire and analyzed using descriptive statistics and non-parametric tests. The overall questionnaire score reflected a moderate level of perceptions toward AI utilization (M = 3.22). Participants reported high agreement regarding the importance and usefulness of AI in administrative processes (M = 3.96). In contrast, the ethical considerations dimension yielded a low score (M = 2.38), indicating limited perceived ethical, legal, and professional acceptability of AI utilization and notable reservations regarding the safeguards and conditions governing its responsible use. Perceptions of AI-supported administrative decision-making were moderate (M = 3.07), suggesting that AI was viewed primarily as a supportive resource rather than as a substitute for professional judgment. No statistically significant differences were found according to professional role, H(2) = 4.286, p = 0.117, or years of experience, H(2) = 2.249, p = 0.325. Significant differences were found according to AI-related training, H(2) = 9.282, p = 0.010. Holm-adjusted pairwise comparisons indicated that participants who had completed three to five courses and those who had completed more than five courses obtained significantly higher overall scores than those who had completed fewer than three courses, whereas no statistically significant difference was found between the two higher-training groups. The findings highlight the importance of targeted professional development, ethical safeguards, regulatory clarity, and continued human oversight in the responsible use of AI in special education administration.