An Inclusive Teacher-Centred Platform for AI-Supported Adaptive Learning in Neurodivergent Children

Abstract

Neurodivergent children in elementary school, specifically those with Autism Spectrum Disorder (ASD) and Attention Deficit Hyperactivity Disorder (ADHD), require learning tasks tailored to their sensory and cognitive profiles. However, educators face significant workloads when personalising materials, often hindered by digital tools lacking transparency and professional control. This paper proposes an inclusive, teacher-centred platform featuring a gamified assistant for adaptive learning. Its core is a controlled AI pipeline that transforms professional materials into safe, traceable activities.A participatory focus group with teachers, therapists, and neurodivergent adults established six foundational design recommendations mapped to safety and human diversity. To ensure safety and agency, the architecture needs an AI-output validation via a Human-in-the-Loop (HITL) dashboard and active source grounding. To prioritise emotional safety, traditional timers should be replaced with progress-based temporal anchors, while sensory safety is enhanced through a personalised audio environment integrating teacher voice recordings. Additionally, native sensory toggles and non-destructive navigation mitigate cognitive fatigue. This work contributes a transparent framework for adaptive task generation, addressing opacity and loss of professional agency in AI-supported special education.

Keywords: Neurodivergent children, Inclusive design, Adaptive learning, Virtual assistant

DOI: 10.54941/ahfe1008196