The Effectiveness of AI-Enabled and AI-Generated Games for Language Development in Children with Autism Spectrum Disorder: A Systematic Review

Abstract

Spoken-language and communication difficulties are core clinical features of autism spectrum disorder (ASD). However, the efficacy of AI-enabled or AI-generated games in advancing language development for autistic children remains unclear. This systematic review examined seven databases up to March 30, 2026. Studies were rigorously screened using PICOS criteria. Methodological quality and evidence certainty were assessed using ROBINS-I, JBI, and GRADE frameworks. Six reports involving approximately 131 participants met the inclusion criteria. Engagement and acceptability for these AI games were consistently high (low certainty of evidence). Direct language improvements emerged in only one generative-AI trial. These dose-dependent gains were limited to a small, high-exposure subgroup (very low certainty). Furthermore, the single controlled study demonstrated no significant advantage over treatment as usual. AI-supported games for language development in ASD show promise but remain fundamentally untested. Future research urgently requires adequately powered randomized controlled trials. These trials must utilize validated language measures and report safety outcomes transparently.