SLxAI Summit: Building Disability-Inclusive AI

The SLxAI Summit convened in Boston this week, bringing together researchers, advocates, and technologists to address the critical gap in AI accessibility for deaf and hard-of-hearing communities, establishing 'Deaf-Safe' AI design principles.

Key highlights: • The summit focused on sign language AI development, identifying significant gaps in current large language and vision models for ASL and other sign languages. • Key concern: most AI assistants — including voice-based models — are designed for hearing users by default, creating systemic exclusion for the deaf community. • 'Deaf-Safe' principles proposed: AI systems should never assume voice as the primary interface, should support visual-spatial communication natively, and should include deaf community representatives in development. • As AI replaces more human-to-human services (customer support, healthcare triage, education), the exclusion risk for non-auditory users grows proportionally.

Why it matters: Accessibility is not a feature add-on — it is a design prerequisite. As AI becomes the interface layer for essential services, exclusion of any community becomes a rights and equity issue.