AI in Special Education: Balancing Technical Potential with Ethical and Practical Realities

As artificial intelligence (AI) becomes more prominent in education, its role in special education is increasingly defined by the tension between technological potential and implementation realities. This narrative review synthesizes 13 studies published between 2022 and 2025 to examine how recent scholarship positions AI and large language models (LLMs) in inclusive and special education. The review identifies a growing shift toward AI-based identification of neurodevelopmental disorders through pattern recognition and digital biomarkers. It also highlights real time personalization, with AI adapting curriculum delivery to students' cognitive and emotional needs to support diverse learners. In addition, AI shows promise in reducing teacher workload by automating administrative tasks, including the drafting and monitoring of Individualized Education Programs (IEPs). At the same time, the literature raises concerns about pedagogical displacement, algorithmic bias, and the tension between data surveillance and cognitive liberty. These ethical issues are compounded by practical limitations, particularly the lack of longitudinal evidence and the geographic imbalance in existing research, which may deepen inequities in low resource settings. The review suggests that AI should function as a multidisciplinary support tool that preserves the professional judgment and relational foundations of human-centered pedagogy.