AI in Teaching and Learning for Autistic Students: Evidence, Limitations, and Instructor Decision Points

Abstract

Artificial intelligence (AI) is increasingly present in teaching and learning, yet research is emerging so quickly that instructors risk overgeneralizing tool-specific gains into broad claims that may not hold across learners or settings. This poster surveys the field of AI-enhanced educational tools with a focus on autistic students and a neurodiversity-affirming framing that treats differences as variation rather than deficit. Using an integrative literature review approach, we synthesize recent peer-reviewed studies and reviews (primarily 2021–present) across major AI tool types relevant to instructional contexts, including intelligent tutoring and adaptive learning systems, robot-mediated and embodied-agent interventions, and emerging generative AI supports. Across studies, AI tools show promising associations with targeted communication outcomes (for example, increased verbal initiation and turn-taking in structured learning environments) and potential benefits for personalization and comprehension when adaptive or generative features are used intentionally. However, the evidence base remains uneven, with recurring limitations such as small samples, short intervention windows, heterogeneous measures, and limited longitudinal, classroom-embedded evaluations, especially for newer generative AI approaches. The poster concludes with an instructor-oriented set of decision points for adopting AI in teaching and learning with autistic students, emphasizing ethical and implementation considerations. It includes questions about when personalization features genuinely serve students' needs versus when they shift instructional judgment to AI systems, as well as privacy, bias, transparency, and impacts on teacher workload. The goal is to support educators in making intentional, evidence-informed choices about when and how AI strengthens learning and student experience.

Presenters