Iterative Negotiation in Intent-Based Interactions: Bridging the User Interface Accessibility and Usability Gap with LLMs

Despite widespread digital technology, many systems remain inaccessible to individuals with intellectual disabilities (ID). While Large Language Models (LLMs) offer potential to improve access, they struggle with ambiguous intents, conflicting preferences, and response reliability. Existing prompt and interface designs often overlook the linguistic, cognitive, and perceptual challenges of users with ID, further limiting usability and safety. The research will adopt Research through Design and a participatory approach to investigate how negotiation strategies and multimodal interaction can be effectively integrated into LLM-driven conversational agents to enable intent-based multimodal interaction and enhance system accessibility and usability. Expected outcomes include an iterative negotiation framework supporting personalized human-AI collaboration for people with ID, along with inclusive interface design guidelines, toolkits, and system architecture requirements to promote digital well-being.