Artificial Intelligence in Special Education: A Review of Adaptive Learning Systems, Personalized Support, and Ethical Challenges for Learners with Disabilities

Abstract

Artificial intelligence (AI) is increasingly entering special and inclusive education through adaptive learning platforms, intelligent tutoring systems, learning analytics, speech and language tools, computer vision, generative AI, and assistive technologies. For learners with disabilities, these systems can reduce barriers in curriculum access, communication, formative feedback, assessment participation, independence, and teacher workload. However, the evidence base remains uneven. Current reviews show stronger concentration on dyslexia, adaptive instruction, and selected intelligent tutoring applications, while comparatively fewer studies address dyscalculia, intellectual disability, complex communication needs, multiple disabilities, low-resource settings, long-term classroom implementation, or learner agency. This critical review synthesizes research and policy literature on AI in special education across three interlinked dimensions: adaptive learning systems, personalized support for diverse disability groups, and ethical governance. The review argues that AI should not be understood as a replacement for special educators, therapists, or individualized education planning. Its strongest role is as a carefully governed support layer that can improve diagnosis-informed instruction, multimodal access, formative feedback, communication, classroom decision-making, and inclusive participation. The paper contributes a disability-responsive AI architecture, a risk-to-governance matrix, and a practical adoption framework for schools. Responsible AI in special education must be accessible by design, pedagogically valid, disability-responsive, transparent, privacy-protective, bias-audited, teacher-supervised, and aligned with learner dignity and self-determination.