Closing the Digital Gender Divide Before AI Makes It Permanent

Every time a Caribbean government official speaks about the digital gender divide, the conversation gravitates toward coverage maps and mobile penetration rates. More towers. Better broadband. Subsidised data plans. These are real needs, and they matter. But they are not the fundamental problem, and treating them as if they are has allowed a governance failure to persist under the cover of an infrastructure narrative.

My hypothesis is this: the Caribbean digital gender divide is not primarily a connectivity problem but a governance failure, and without deliberate, structured policy intervention, AI systems being adopted across the region's economy will entrench these inequalities into the economic infrastructure for a generation. Once AI systems are trained, deployed, and normalised, the data gaps that produced their biases become invisible. The discrimination becomes architectural.

What follows sets out the evidence for this claim, names the governance mechanisms that have failed, and describes what deliberate policy intervention actually requires, as distinct from well-intentioned but unfocused digital inclusion rhetoric.

What the Data Actually Shows

The GSMA Connected Women Commitment initiative and its associated annual reporting have tracked the mobile gender gap across developing regions for years. The GSMA Connected Women Report 2024 confirms that women in low- and middle-income countries remain markedly less likely than men to own a mobile phone and to use mobile internet. The gap has narrowed in some markets but remains persistent, and in several regions it has not meaningfully changed in three years.

The barriers the GSMA data consistently identifies are not primarily about infrastructure. They are about affordability, literacy, social norms around women's technology use, safety and harassment online, and the perceived relevance of digital content to women's lives. These are social and policy problems. They respond to social and policy interventions: subsidised devices targeted at women, digital literacy programmes designed for women's contexts, regulatory action against online harassment, and content development that speaks to women's economic and social priorities.

Caribbean-specific data is thinner than it should be, which is itself a governance failure I will return to. But the ITU's ongoing reporting on gender and the digital divide confirms the pattern holds across developing regions, and the OECS has produced analysis showing gender disparities in digital access and digital skills across Eastern Caribbean member states. The gap is real, documented, and persistent despite years of digital inclusion commitments.

The Intersecting Barriers Caribbean Women Face

Caribbean women do not face a single barrier to digital inclusion; they face a constellation of reinforcing disadvantages. Gender intersects with race, class, geography, age, and disability to produce patterns of exclusion that vary across the region and within individual countries. A rural woman in St. Elizabeth, Jamaica, faces different barriers than a woman in Kingston, and both face different barriers than a woman in a small island developing state with limited telecommunications infrastructure.

Affordability is a consistent factor. Mobile data plans consume a higher proportion of income for women than for men, because women's incomes are lower on average. The gender pay gap is documented across CARICOM labour markets by the ILO and by national statistics offices. When data costs represent a large share of weekly income, women make rational decisions to limit consumption, and these rational decisions accumulate into a usage pattern that widens the digital divide.

Online safety is a second consistent factor that infrastructure investment does not address. Women across the Caribbean report experiences of harassment, surveillance, and abuse in online spaces. The absence of clear online safety legislation in most CARICOM jurisdictions means that the internet is a more hostile environment for women than for men, and women's decisions to limit their online presence are therefore not failures of digital literacy but rational safety calculations.

Caribbean women's educational attainment at tertiary level exceeds men's by a wide margin. The argument that women lack the capacity to engage with digital technology is empirically unsustainable. What they lack is safe, affordable, relevant access to digital infrastructure, and the policy environment that would create it.

How AI Converts a Gap into a Permanent Structure

The digital gender divide would be a serious problem without AI. With AI, it can become a generational one, and the mechanism is worth setting out plainly.

AI systems learn from data. The quality of their outputs for any given population depends on the quality and quantity of data that population has contributed to training. Women who are less digitally connected, who use fewer digital services, who transact less frequently through digital financial systems, and who are less represented in digital health and education platforms produce less data. Less data means poorer AI performance for those populations.

AI-driven financial services will be less accurate for women who have thinner digital transaction histories. AI-assisted healthcare tools will perform worse for women whose health experiences are underrepresented in training datasets. Agricultural AI advisory services will be less useful for female farmers, who are underrepresented in the farmer registration databases from which training data is drawn. Government service delivery AI will make more errors for women, whose interactions with formal government systems have historically been mediated by male household heads in ways that create incomplete individual records.

Each of these accuracy gaps, when embedded in production AI systems, becomes a self-reinforcing dynamic. Women receive worse service, trust AI systems less, engage with them less, contribute even less data, and the next generation of AI systems is trained on an even more gender-skewed dataset. The gap widens with each iteration.

The UN High-level Advisory Body on AI, which published its final report "Governing AI for Humanity" in September 2024, explicitly identifies equitable AI governance as a priority and calls for developing country participation in international standard-setting. But participation in international forums does not automatically translate into domestic policy that prevents this accumulating AI disadvantage for women.

The Governance Failures That Created This Problem

I want to be direct about where governance has failed. Digital inclusion policy in the Caribbean has been characterised by aspirational frameworks without implementation mechanisms, gender commitments without gender-disaggregated targets, and technology investment without equity analysis.

The CARICOM Regional Digital Economy Policy Framework establishes a regional vision for inclusive digital economies. But the Framework has not been operationalised with specific gender equity targets, funded implementation plans, or accountability mechanisms for member states. The gap between the Framework's principles and the lived experience of women in CARICOM member states is not small.

Jamaica's Vision 2030 includes commitments to gender equality and to digital economy development. These commitments exist in different sections of the plan and have not been integrated. The question of how Jamaica's digital economy development will deliver gender-equitable outcomes has not been systematically addressed in national development planning. This is a governance failure: not a failure of intent, but a failure of institutional design.

The G20 Brazil summit in November 2024 produced further commitments on digital inclusion and AI governance, including attention to gender equity. Caribbean states that are not G20 members are not party to these commitments, but they are subject to the AI systems and governance norms that G20 members produce. Governance that is made elsewhere, without Caribbean input, will not reflect Caribbean women's circumstances.

What the EU AI Act Signals for Caribbean Policy

The EU AI Act entered into force on 1 August 2024. Its provisions on high-risk AI systems, transparency, and prohibited practices set a benchmark for AI governance that Caribbean states should study seriously, not because they are legally bound by it, but because the tools being deployed in Caribbean markets are largely produced by entities subject to it.

A European-made AI system deployed in healthcare in Jamaica will have been assessed against EU standards before it reaches the Caribbean market. But the specific impacts of that system on Jamaican women, particularly those with limited digital footprints or those whose health data is underrepresented in European training datasets, will not have been assessed at all. EU compliance does not equal Caribbean equity.

The Seoul AI Safety Summit of May 2024 produced international commitments to AI safety that implicitly include equity considerations. Caribbean governments should treat their participation in these international conversations as a mandate to develop domestic frameworks, not as a substitute for them.

Recommendations

1. Establish gender-disaggregated digital inclusion targets in all national digital economy plans. Every CARICOM member state with a digital economy strategy should add specific, measurable targets for closing the gender digital divide within defined timeframes. Targets should cover mobile internet access, digital skills, digital financial inclusion, and participation in the digital economy workforce, reported annually to a designated accountability body.
2. Require gender impact assessments before deploying AI systems in public services. Any government ministry or public authority deploying an AI system for service delivery should be required to conduct and publish a gender impact assessment before deployment. The assessment should address how the system will perform for women with limited digital histories, and what remediation is in place when performance gaps are identified.
3. Fund Caribbean-specific digital safety legislation addressing online harassment. The absence of dedicated online safety law in most CARICOM jurisdictions is a documented barrier to women's digital participation. Legislating against online harassment, with criminal penalties and accessible civil remedies, is a prerequisite for meaningful digital inclusion. Model legislation developed at the CARICOM level, with member state implementation supported by technical assistance, would be the most efficient approach.
4. Integrate gender equity analysis into the CARICOM Regional Digital Economy Policy Framework's implementation structures. The Framework should be amended to include a gender equity review mechanism, with annual reporting by member states on gender-disaggregated indicators of digital inclusion. A dedicated technical working group on gender and the digital economy, with UN Women Caribbean as a partner, should oversee this process.
5. Invest in Caribbean-specific data collection on women's digital access and use. The policy gap in gender and digital data across the Caribbean limits the quality of both policy design and AI training. The Statistical Institute of Jamaica and its CARICOM equivalents should incorporate mandatory gender-disaggregated digital access and use questions into regular household and labour force surveys, and publish results annually in a form accessible to AI researchers and policymakers.
6. Create a CARICOM AI equity watchdog with gender expertise. A regional body with authority to review AI deployments in CARICOM member states for equity impacts, including gender impacts, would provide the accountability mechanism the current framework lacks. This body should include technical AI expertise, gender expertise, and legal expertise, and should have the power to recommend suspension of AI deployments that fail equity standards.

Conclusion

The digital gender divide in the Caribbean is not waiting for a policy response. It is producing consequences every day, in the labour market opportunities women cannot access, the financial services they cannot use, the healthcare information they cannot obtain, and the AI-powered systems that perform less well for them than for their male peers.

The AI transition that is underway across the region's economy is not a neutral technological upgrade. It is a process that will either widen or narrow these existing inequalities, depending entirely on whether governance frameworks require equitable design and equitable outcomes. At present, those frameworks are absent.

The window for preventing AI from making the digital gender divide permanent is measured in years, not decades. The infrastructure choices, the data collection decisions, the AI procurement standards, and the regulatory frameworks adopted in the next three to five years will determine whether Caribbean women participate in the AI economy on equal terms or are written out of it by technical systems that were never designed to include them. Governance must act before the architecture sets.

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. She conducts legal and regulatory framework reviews and develops policy recommendations for legal reform in AI governance, data protection, human rights, and gender equality. Dr. Budall has knowledge of international and regional AI governance standards and has advised Caribbean government institutions and regional organisations on inclusive AI policy. She is a researcher and consultant working across the CARICOM region on digital economy governance, women's rights in the digital age, and equitable technology development. Contact: insights@starapple.ai