This Special Thematic Session (STS) for ICCHP 2026 proposal focuses on Generative AI (GenAI) for cognitive accessibility, addressing how GenAI systems can be designed and evaluated to be accessible, understandable, ethical and useful for people with cognitive impairments.
The session will bring together HCI and NLP research and practice on: (i) accessible GenAI tools for people with cognitive impairments; (ii) LLM-based and other approaches for plain language and easy-to-read generation and adaptation with human-in-the-loop methods; and (iii) Intelligent User Interfaces (IUI) approaches to
reduce cognitive load and improve comprehension.
Applied work is encouraged across domains. The overall goal is to bring together researchers and practitioners to consolidate methods, datasets, metrics and design guidelines that advance cognitively accessible GenAI in real-world settings.
We welcome papers from researchers and practitioners in HCI, NLP, accessibility and assistive technologies. We also encourage contributions from professionals working closely with end users, caregivers and other stakeholders. The session also welcomes foundational and exploratory work that informs the design of future GenAI-based accessible systems.
This STS is organised within the HumanAI project and jointly led by the HULAT research group (Universidad Carlos III de Madrid, UC3M) and the NIL research group (Universidad Complutense de Madrid, UCM).
Topics of interest:
● Accessible, usable and understandable GenAI interaction for people with cognitive impairments
● Plain Language and Easy-to-Read content generation, adaptation and evaluation using Large Language Models (LLMs)
● Human-in-the-loop approaches, participatory design and co-creation methods for GenAI accessibility
● Intelligent User Interfaces (IUI) for cognitive accessibility, including adaptive interfaces, personalization and assistive interaction
● Multimodal GenAI for cognitive accessibility: text–image–audio interaction, pictograms, visual supports and alternative representations
● Cognitive load reduction in GenAI systems: UX patterns, explainability, transparency, trust and user control
● Accessible and inclusive communication: user-oriented language and caregiver support
● Evaluation protocols, datasets and methodologies with target users, including qualitative, quantitative and mixed-methods approaches
● Ethics, risk assessment, bias and safeguards in GenAI systems for cognitive accessibility and vulnerable populations
● Datasets, benchmarks and metrics for evaluating readability, fluency, user preferences, and multilingual considerations
Accepted scientific contributions to this STS will be published in the ICCHP proceedings, following the conference publication routes (Springer LNCS or the ICCHP Open Access Compendium), according to ICCHP rules for Special Thematic Sessions. In addition, the organizing team plans to coordinate a journal special issue to further develop selected contributions beyond the proceedings.
Chairs
Virginia Francisco, Universidad Complutense de Madrid (UCM) – HumanAI / NIL
Rémi Cardon, Universidad Carlos III de Madrid (UC3M) – HumanAI / HULAT
Special Session organised within the framework of Grant PID2023-148577OB (Human-Centered AI: Towards Accessible and Customizable Language Models (HumanAI), funded by MICIU/AEI/10.13039/501100011033 and by FEDER/UE.