Channelling, Coordinating, Collaborating: A Three-Layer Framework for Disability-Centered Human-Agent Collaboration

Why Read This from an HCI Perspective

This piece is highly meaningful for both HCI/UX practitioners and researchers because it reframes AI accessibility not as a ‘personal assistive tool,’ but as ‘collaboration infrastructure.’ In particular, by breaking down design challenges into three layers—information access, task coordination, and co-creation—it provides a concrete analytical framework for judging how far AI should be involved in real team environments working alongside people with disabilities.

CIT's Commentary

The most interesting aspect of this article is that it treats accessibility not as a problem of ‘accurate output,’ but as a matter of reallocating collaboration costs. The distinctions among Channelling, Coordinating, and Co-Creating show how LLM-based systems must go beyond simple summarization or supportive responses to handle work status, roles, and even creative permissions. In disability contexts, it’s especially important to recognize that reducing information asymmetry alone is not enough—you also need to design who makes decisions and who approves edits. However, because triadic collaboration is heavily shaped by power dynamics and organizational norms, in real-world application the key validation points will likely be designing role transparency, error recovery, and the possibility of refusal.

Questions to Consider While Reading

• Q.In Channelling, who should define the criteria for ‘functionally equivalent access,’ and through what procedures?
• Q.In the Co-Creating stage, how can we measure user control when deciding whether an AI’s suggestions count as ‘help’ versus ‘participation’?
• Q.In situations where multiple disability types and organizational power dynamics intersect, in what order might these three layers be likely to break down?