Other

Artificial Intelligence for Inclusive Early Childhood Education: A Systematic Literature Review of Learning, Assessment, and Intervention Technologies

Habibi MM, Islami JMM · 28 Jul 2026

Abstract

Abstract Despite some apprehensions and skepticism, the rapid diffusion of artificial intelligence into educational ecosystems is beginning to change how early childhood education supports, assesses, and intervenes with our youngest children – particularly those at risk for or showing signs of developmental disabilities. This systematic literature review synthesises empirical evidence published between 2020 and 2026 about the design, deployment and evaluation of AI technologies for inclusive early childhood settings. Only those records that employed the PRISMA 2020 framework were searched, outlining a Boolean strategy (i.e., keywords related to AI; developmental education^23; and inclusive education) which resulted in six databases being searched (Scopus, Web of Science, ERIC, IEEE Xplore, SpringerLink and ScienceDirect). After quality appraisal by the Mixed Methods Appraisal Tool and Critical Appraisal Skills Programme checklist, 38 studies were included. Results are structured by the following seven research questions regarding: the AI technologies utilized, learning support, assessment, intervention (including adaptive instruction), classroom management and family engagement/experience; educational outcomes; technological and ethical challenges identified; and requirements for sustainable and equitable implementation (conceptual/infrastructural). The review brings to the surface four overlapping domains: AI-powered adaptive learning systems and intelligent tutoring for individualization; computer vision and multimodal analytics for assessment and behavior recognition; socially assistive robots and large language model-based conversational agents targeting social and language intervention, respectively; and learning analytics dashboards to enhance teacher decision making as well as parental engagement. ImplicationsThe synthesis showed the small benefits of engagement, scaffolded instruction and early identification are mostly consistent from 2012–2023 while concerns around algorithmic bias, data privacy, developmental appropriateness, teacher preparedness also remain padding out. The review propels a Human–AI Collaboration Ecosystem framework with four pillars—pedagogical orchestration, ethical governance, equitable infrastructure and participatory co-design—towards an integrated model for early childhood inclusion of AI. The discussion touches on policy, preparation of teachers and future research.