Large Language Models as Cognitive Support Systems for Neurodivergent Individuals

Abstract

As a cognitive extension of human reasoning, communication, and task management, the use of Large Language Models (LLMs) is being investigated more and more. This paper focuses on their possible role to assist neurodivergent people in three ways: by making cognitive function more accessible, making communication clearer, and by lessening executive functioning demands. Drawing from cognitive science and human–computer interaction, the research views LLMs as malleable instruments that can facilitate the planning, language, memorization, and social interaction of their users. The discussion focuses on how these systems can be incorporated into educational, occupational and assistive environments, to foster inclusion and cognitive equity. It also examines architectural strategies for personalisation, adaptive prompting and context-aware support and the potential dangers of relying too heavily, misunderstanding intent and accessibility imbalance. In summary, LLMs can be considered potential allies in neurodivergent support networks and are not just assistive tools, as they may help to increase autonomy and participation in more complex cognitive activities.