DOCTORAL CANDIDATE
Maja Jønck Hjuler

SUPERVISORS
Prof. Didier Schwab, Université Grenoble Alpes (UGA)
Benjamin Lecouteux, Université Grenoble Alpes (UGA)
Assoc. Prof. Laurianne Sitbon, Queensland University of Technology (QUT)
Jessica Korte, Queensland University of Technology (QUT)

Informatics, Computer Science, Augmented and Alternative Communication, Speech and Natural Language Processing, Human-Computer Interaction

This project explores the potential of emerging Artificial Intelligence (AI) models to enhance Augmentative and Alternative Communication (AAC) methods tailored to the unique competencies, needs, and preferences of individuals with intellectual disabilities. We aim to investigate how advanced text-based AI can support and be enriched by AAC methodologies, fostering a reciprocal relationship that drives innovation in both domains. The project will develop multimodal, multilingual models integrating speech, text, images, pictograms, and other communicative symbols using Large Language Models. Building on insights from research conducted by the Grenoble Informatics Laboratory, this project seeks to advance AAC models that are culturally adaptable, contextually relevant, and responsive to individual users. A key objective is to design personalized communication aids, expanding AAC applications in real-world settings.

The Grenoble-based research team brings expertise in developing language models for written and spoken communication, leveraging France’s advanced GPU computing resources. The project collaborates with Queensland University of Technology in Brisbane, which has extensive experience creating flexible and configurable prototypes for individuals with intellectual disabilities. A key focus is ensuring neurodiverse users are involved at every research and development stage. We aim to design adaptable and customizable AAC prototypes, iteratively refining them through direct participant interaction. Potential AAC applications of this research include the automatic organization of pictogram grids, converting speech or text to pictograms, and generating visual scene displays, thereby broadening the accessibility and effectiveness of communication tools tailored for people with intellectual disabilities.