AI Accessibility for Caribbean People with Disabilities

Disability statistics across the Caribbean are uneven and under-reported, but the consistent finding across the regional surveys is that roughly one in seven Caribbean adults lives with a disability of some kind. Visual, hearing, mobility, cognitive, mental-health, and chronic-illness-related impairments are all part of the population picture. Caribbean disability policy is shaped by the UN Convention on the Rights of Persons with Disabilities, which every CARICOM member has signed or ratified, and by a varied national legal and institutional patchwork that ranges from well-developed (Jamaica, Barbados, Trinidad and Tobago) to very thin elsewhere.

AI is now both an opportunity and a new exclusion risk for Caribbean people with disabilities. This article is for the regional disability organisations, ministries of social services and labour, employers, schools, and the broader Caribbean public who need to understand which AI tools genuinely help and which quietly raise the barrier.

What AI Is Doing Well for Caribbean Accessibility

Several AI capabilities have direct, useful applications for Caribbean people with disabilities.

Image description for blind and low-vision users. AI-powered image description (the screen-reader "describe image" feature, the standalone apps that narrate photos and surroundings) has become broadly useful. Caribbean blind and low-vision users report that the major tools work in standard English settings; performance on Caribbean-accented English and on creole-language interfaces is less reliable.

Speech-to-text captioning. Real-time captioning, including in classrooms, religious services, and broadcast media, is genuinely useful for deaf and hard-of-hearing Caribbean users, with the language caveats discussed in the creoles article. The Caribbean Council for the Blind and the regional disability organisations have begun to track this.

Text simplification. AI text-simplification tools are useful for users with cognitive disabilities, with low literacy, or for whom standard written English is a second or third language. The Caribbean public-information environment, which often uses formal English heavily, can be made meaningfully more accessible through these tools.

Communication augmentation. AI-supported communication boards and predictive text systems are useful for users with motor disabilities or aphasia. These tools have improved markedly in the past five years.

Mobility and navigation. AI-powered mobility apps (computer vision for obstacle detection, indoor navigation, public-transport assistance) work better in well-mapped urban environments and less well in the Caribbean reality of mixed-use rural and small-town infrastructure.

Where AI Is Quietly Excluding Caribbean People with Disabilities

Five exclusion patterns are worth naming.

Hiring and HR AI. Many of the AI hiring tools entering Caribbean workplaces have been shown, in international research, to disadvantage applicants with disabilities. Video-interview AI penalises non-standard facial expression, atypical eye contact, or speech patterns associated with several disabilities. Resume-screening AI can downgrade career gaps that disabled candidates often have. Caribbean employers adopting these tools without disability-aware testing are quietly excluding qualified disabled candidates.

Benefits and means-testing AI. Automated benefits adjudication systems used in some Caribbean welfare programmes use pattern-matching that can disadvantage disabled claimants whose circumstances are heterogeneous. The Dutch childcare benefits scandal and the Australian Robodebt experience are warnings.

Inaccessible AI interfaces. Some AI tools are themselves inaccessible. Chatbots that do not work with screen readers, voice interfaces with no text alternative, image-only generative tools, captcha walls that defeat assistive technology. Caribbean organisations procuring these tools without accessibility checks are creating accessibility regressions.

Foreign-trained accessibility tools. Some AI accessibility tools were trained on data that does not reflect Caribbean users. Sign-language AI tools trained on ASL or BSL may not handle Jamaican Sign Language (JSL), Trinidadian Sign Language, or the regional sign languages of the OECS, Suriname, and Haiti. The tool that works for a deaf user in New York may not work for the deaf user in Kingston using JSL.

Cost barriers. Several of the best AI accessibility tools are paid subscriptions. Caribbean disabled users, who face income gaps that are well-documented across the region, often cannot afford them. Free alternatives exist but are often inferior. AI is, in this sense, recreating the income-based accessibility gap that earlier assistive technology had begun to narrow.

What Caribbean Disability Organisations Should Do

Four priorities.

Document the AI accessibility experience of members. The regional disability organisations have credibility with the policy process that no other body can match. Documented evidence of AI tools failing Caribbean disabled users is what moves regulatory and employer practice.

Engage with regional sign-language and Caribbean-creole AI tool development. The University of the West Indies, the regional teacher-training colleges, and the deaf-led organisations across CARICOM have the linguistic and pedagogic expertise that any credible Caribbean sign-language AI tool will require.

Push for accessibility procurement standards. Caribbean ministries and large employers respond to clear, organisation-backed procurement expectations. A regional accessibility procurement checklist for AI tools is achievable and would have material effect.

Participate in regional AI policy processes. Disability voices are usually under-represented in technology policy. The disability organisations that show up to CAIRMC consultations, CARICOM technology dialogues, and national digital-strategy processes will, in time, shape the outcomes.

What Caribbean Employers Should Do

Three steps for any Caribbean employer using AI in hiring, performance management, or workplace tools.

Audit your AI hiring stack for disability disparate impact. Even a small-sample audit, conducted by HR with disability-organisation input, reveals patterns the vendor's marketing does not.

Offer reasonable accommodation around AI tools. A candidate who declines a video-interview AI in favour of a phone or in-person interview should be accommodated without penalty. The right to such accommodation is, in many Caribbean jurisdictions, already enforceable.

Verify accessibility of internal AI tools. Before deploying an AI tool internally (chatbot, scheduling assistant, document-summary tool), check that it works with screen readers, with keyboard-only operation, and with the assistive technology that disabled employees actually use.

What Caribbean Schools and Universities Should Do

Three steps.

Treat AI accessibility as a teaching and learning priority. AI tools that help disabled students should be made available; AI tools that exclude them should be avoided or accommodated around.

Adapt assessment practice. Some AI-detection approaches in academic settings penalise disabled students whose use of assistive AI is itself a reasonable accommodation. The assessment policy needs to distinguish between AI use that is part of accommodation and AI use that substitutes for the student's own work.

Train educators. Teachers and lecturers across the region need a working understanding of which AI tools support disabled learners and which obstruct them. The teacher-training colleges and the university faculties of education are the right place to build this.

What Caribbean Governments Should Do

Three priorities.

Make the public-sector digital footprint accessible. Caribbean government websites, online forms, and e-services that incorporate AI should meet the existing accessibility standards (WCAG, EN 301 549 where adopted). Several Caribbean governments have made progress; many have not.

Build disability impact assessment into procurement. Where a Caribbean ministry is procuring an AI tool for use on or by citizens, a disability impact assessment should be part of the evaluation. This does not require new law in most jurisdictions; it requires updating procurement guidance.

Resource the regional sign-language and accessibility-AI work. A modest, sustained regional investment in Caribbean sign-language datasets, Caribbean-creole accessibility tools, and disability-aware AI evaluation benchmarks would, over a decade, change the regional accessibility landscape.

Frequently Asked Questions

Is the Caribbean's existing legal framework on disability enough?

Several Caribbean jurisdictions have disability legislation. The UN Convention is binding on all CARICOM signatories. The legal floor is uneven; the enforcement is uneven; the AI-specific guidance is generally absent. The existing legal framework is the right starting point but needs to be supplemented with AI-specific guidance from disability commissioners, equality bodies, and labour regulators.

Are AI tools accessible by default?

No. Accessibility has to be designed in or retrofitted. AI tools built without accessibility consideration will, in practice, be partially or wholly inaccessible to many disabled users. The Caribbean organisations procuring these tools are the ones that can change that incentive.

What about Caribbean sign languages specifically?

Caribbean sign languages, including Jamaican Sign Language and the various OECS, Suriname, Haitian, and other regional varieties, are under-represented in AI sign-language tools. The position is similar to the spoken-creole position covered in the language sovereignty article. Speaker-led, consent-based data collection through the deaf-led organisations is the right path.

Can a small Caribbean employer afford accessibility-AI tooling?

Most accessibility-AI tools that matter (screen readers, captioning, image description) are available at low or no cost on standard operating systems and mobile devices. The cost question is more about training, accommodation, and procurement discipline than about expensive new tools.

How does this connect to CAIRMC's broader AI risk work?

Accessibility is part of AI risk. A tool that produces sound outputs for the average user and exclusion outputs for disabled users is producing risk that is concentrated in a vulnerable population. The same risk management practices that CAIRMC publishes for financial-services AI apply, with adjustments, to accessibility.

The Caribbean Accessibility Bottom Line

Caribbean people with disabilities have spent decades building the accommodations, advocacy, and infrastructure that produces the partial inclusion the region has today. AI is now reshaping that landscape in ways that are not, on balance, accidental. The tools that include disabled users do so because someone insisted that they should; the tools that exclude do so because no one in the procurement process knew or asked. The Caribbean institutions that build disability voice into their AI procurement and policy decisions will produce systems that work for everyone. The ones that do not will discover, slowly and at considerable human cost, that "accessibility" is what is needed to make AI work for the population, not an add-on for a minority.