Designing Equitable AI Systems for the Caribbean: A Policy Framework

There is a pattern in how governments respond to gender equity critiques of technology policy: they add a paragraph. Sometimes they add a clause. On a good day, they commission a study. What they rarely do is go back to the design assumptions embedded in the policy itself and ask whether those assumptions were ever consistent with gender equity in the first place.

I argue that this pattern is the central obstacle to equitable AI in the Caribbean. My hypothesis is that equity in AI cannot be achieved by appending gender language to frameworks built around different assumptions. It requires a distinct policy architecture that places the lived experience of Caribbean women, particularly those facing intersecting disadvantages of gender, race, class, and geography, at the centre of AI system design rather than treating them as an afterthought to be accommodated in a footnote.

Why the Existing Approach Is Insufficient

Caribbean technology policy, like technology policy in most of the world, was designed primarily by people who were not thinking about gender. The CARICOM Regional Digital Economy Policy Framework, various national ICT policies, and emerging AI strategy documents across the region share a common structural characteristic: they treat equity as a value to be affirmed rather than a design requirement to be specified.

The EU AI Act, whose prohibition provisions became effective on 2 February 2025, takes a categorically different approach. It does not merely affirm non-discrimination as a value; it specifies legal requirements, risk classifications, technical standards, and enforcement mechanisms that operationalise non-discrimination as a legal obligation.

UNESCO's "Guidance for Generative AI in Education and Research", published in September 2023, demonstrates what sector-specific application of equity principles looks like. The Guidance addresses curriculum representation, assessment fairness, data privacy for students, and the specific risks that generative AI poses for students from disadvantaged backgrounds.

The Core Principles of an Equitable AI Framework

Representational adequacy. AI systems deployed in the Caribbean must demonstrate that their training data adequately represents Caribbean women, including women in rural areas, women in informal employment, and women from marginalised communities.

Participatory design. Caribbean women must be involved in the design and governance of AI systems that affect them, not as consultees at the end of a design process but as participants in defining the problem the AI system is meant to solve.

Transparency with comprehensibility. Transparency that produces 200-page technical documentation is not meaningfully accessible to a woman in Kingston who wants to understand why an AI system denied her a bank loan. Comprehensible transparency, in plain language and in accessible formats, must be specified as an explicit requirement.

Accountability with remediation. An equitable framework requires not only that harms be identified but that they be remedied. Accountability mechanisms that document harm without correcting it protect institutions rather than individuals.

The Legal Architecture

An equitable AI framework for the Caribbean does not require starting from scratch. The Jamaica Data Protection Act 2020 provides data rights that can be extended to address AI-specific harms. The Employment (Equal Opportunities) Act provides a basis for addressing gender discrimination in AI-mediated hiring. The Consumer Protection Act provides a basis for addressing unfair practices in AI-driven commercial services.

At the regional level, the CARICOM Regional Digital Economy Policy Framework should be supplemented by a binding regional AI Protocol that establishes minimum standards for equitable AI deployment across member states.

The Convention on the Elimination of All Forms of Discrimination Against Women (CEDAW), to which all CARICOM member states are parties, creates positive obligations to eliminate discrimination in employment, education, health, and economic life. Caribbean states should report on AI-related gender discrimination risks in their periodic reviews.

Sector-Specific Application

In financial services, AI-driven credit scoring and financial product delivery are already operational across the Caribbean. The Jamaican banking sector and its equivalents across CARICOM should be required to conduct gender equity audits of any AI system used in credit assessment.

In healthcare, AI diagnostic tools, triage systems, and health information platforms are entering Caribbean health systems. Caribbean Ministries of Health should require health AI vendors to provide evidence of validation studies in demographically comparable populations before public procurement approval.

In education, Caribbean education ministries should conduct an annual review of AI tools in use in schools and tertiary institutions, assessing each tool against UNESCO's Guidance criteria, with particular attention to whether the tool performs equally for students from different socioeconomic and gender backgrounds.

Recommendations

1. Enact AI-specific legislation in Jamaica within 24 months. The legislation should establish mandatory bias audit requirements for high-risk AI systems, a right to explanation for automated decisions, mandatory gender impact assessments for public sector AI procurement, and regulatory authority for the Office of the Information Commissioner.
2. Negotiate a binding CARICOM AI Protocol establishing minimum equity standards. Market access conditions should be attached: systems not meeting the Protocol's minimum standards cannot be deployed across the CARICOM single market.
3. Require gender equity audits as a condition of public sector AI procurement. Any government ministry procuring an AI system should be required to obtain an independent gender equity audit before contract signature.
4. Fund participatory design processes for AI systems in healthcare and education. These consultations should be funded, not voluntary, and their outcomes should be documented and reflected in system requirements.
5. Apply CEDAW reporting mechanisms to AI-related gender discrimination. Caribbean states should include AI-related gender discrimination risks in their next CEDAW periodic reviews.
6. Establish a Caribbean AI equity fellowship programme for women. A funded fellowship programme, placing women with gender expertise in AI regulatory roles in Caribbean governments and regional institutions, would begin to build the human infrastructure that equitable AI governance requires.

Conclusion

Equitable AI in the Caribbean will not emerge from goodwill and general principles. It will emerge from specific legal requirements, adequately funded institutions, accountable governance processes, and the deliberate inclusion of Caribbean women's expertise and experience in AI system design and oversight.

An equitable AI framework for the Caribbean is not a luxury or a future priority. It is an urgent necessity, and the time to build it is while AI systems are still being adopted rather than after they have embedded themselves so deeply into the region's economic infrastructure that reform becomes structurally difficult.

About the Author

Dr. Shirley Budall is a Caribbean expert in gender, inclusion, and AI governance with demonstrated experience in the ethical, legal, social and governance dimensions of artificial intelligence and digital technologies. Contact: insights@starapple.ai