Opportunity
AI development processes that actively involve persons with disabilities—through co-design, testing, auditing, and leadership roles—can lead to systems that better reflect a wider range of needs, experiences, and communication styles. A robust involvement can improve the quality, usability, and fairness of AI tools by identifying issues early and shaping approaches based on lived and living experience. Active participation of persons with disabilities in how AI systems are built and governed can also create opportunities for meaningful employment, leadership, and knowledge-sharing.
Limitation
Participation needs to be intentionally structured by organizations so it is meaningful, not symbolic or performative. There is a risk that involvement of persons with disabilities is limited to a small, tokenized group or used in ways that do not influence real decisions. Persons with disabilities are not a single, uniform group, so multiple perspectives must be included to avoid narrow or incomplete design outcomes. This work must be properly resourced, including fair compensation, accessible processes, and sustained support over time rather than one‑time consultations. Ongoing feedback, local engagement, and transparent decision‑making are essential to ensure participation leads to real impact, not just stated commitments.
Illustrative Examples
Disability-led audits of AI systems
Disability‑disaggregated audits led by persons with disabilities and Disabled Persons’ Organizations (DPOs) help identify and prevent discrimination in AI systems. Case studies of audits of admissions and remote proctoring tools show that this approach can improve transparency, produce public reports, and reduce exclusion.
Caution
Audits should be led by disabled‑led organizations, fairly paid, and backed by regulatory enforcement. There is a real risk of audits being used only for appearances (“audit‑washing”) if they are not independent, resourced, and accountable.
Inclusive co-design produces stronger AI for everyone
Designing AI with persons with disabilities—rather than for them—leads to more inclusive and usable tools. When people with disabilities are involved throughout design and testing, their needs are directly considered, and tools can be refined together to better support a wide range of users.
Caution
Inclusive co‑design depends on ongoing, meaningful participation, not one‑time consultation. Persons with disabilities are not a single group; needs, priorities, and lived experiences differ across disabilities and contexts. Co‑design must therefore be continuous, flexible, and responsive, with regular opportunities for feedback, iteration, and shared decision‑making. Local, community‑based co‑design spaces (such as schools, community centres, and community organizations) help ensure participation is grounded in real‑world use and everyday environments. “Nothing about us without us” must be enacted through sustained co‑design practices, shared authority, and accountability—not treated as a symbolic commitment.