Dematerialising disability support: artificial intelligence as a driver of equity and social sustainability in ASEAN classrooms

Purpose This critical narrative review positions artificial intelligence (AI)-enabled “dematerialisation” as a potential structural substitute for materially intensive disability support in resource-constrained ASEAN classrooms, rather than only as a pedagogical support tool. Using the Social Model of Disability and Critical Disability Theory, it asks how dematerialised support works, how it shapes equity and social sustainability, and what governance prevents a new digital divide. Design/methodology/approach A SANRA-aligned critical narrative review was conducted with ROSES-style evidence reporting. Searches (2020–2025) covered Web of Science, Scopus, Semantic Scholar and targeted grey literature. Full texts were coded using theory-led iterative coding, constant comparison and abductive reasoning to link technical mechanisms to inclusion outcomes and governance conditions. 36 studies met the final theory-led inclusion threshold. Findings The synthesis organises the evidence into three modes of dematerialisation. Mode A dematerialises information access through multimodal AI such as speech-to-text, text-to-speech, captioning, translation and computer-vision narration, expanding access where specialised materials are scarce. Mode B dematerialises human intermediation through assistive agency and pedagogical role redesign, shifting teacher workload towards facilitation while AI provides continuity scaffolds. Mode C dematerialises infrastructure burdens via micro-infrastructure such as accessibility-first platforms, AR/VR supports, robotics and adaptive devices that compensate for missing retrofits, but remain vulnerable to connectivity gaps and maintenance failure. Originality/value The review advances a dematerialisation lens and a Policy-R2R framework that links core mechanisms to sustainable equity outcomes in ASEAN. In doing so, it supports progress towards SDG 4 (Quality Education) and SDG 10 (Reduced Inequalities) by outlining how accessible AI systems can be scaled without reproducing a new digital divide.