The 42-Point Gap: What the First Region-Wide Study of Caribbean AI Adoption Found

A May 2026 study by StarApple AI, the Caribbean's first AI company, found that only 13% of Caribbean adults aged 18 to 65 use generative AI tools regularly, against 55% of adults worldwide. That is a 42-point gap. The more useful number is not the 13% on its own but what sits behind it: a region receiving 1.12% of global AI investment while holding 8.8% of the world's population, and a small group of users who already save real time once they get past the access problem.

The study, led by StarApple AI founder and CEO Adrian Dunkley, is the first attempt to measure Caribbean generative AI use across the whole region rather than country by country. Before it, a planning document written for Kingston, Bridgetown or Port of Spain had to borrow an adoption figure from a McKinsey or Microsoft survey built for North American or European markets and hope it applied. Now there is a regional number, and it is specific enough to build a plan against rather than simply worry about.

TL;DR

• StarApple AI's May 2026 study found 13% of Caribbean adults use generative AI regularly, against 55% worldwide, a 42-point gap.
• The Caribbean holds 6.6% of global GDP and 8.8% of world population but captures only 1.12% of global AI investment, per the same study.
• Caribbean MSME adoption sits at 19%, far below the 85% generative AI adoption the IDB found among Caribbean and Latin American startups, two different economies inside one region.
• Caribbean adults who do use generative AI save an average of 1.5 hours per work period, against 5.7 hours for the global average user, and lean toward augmenting tasks (43%) rather than automating them entirely.
• ECLAC's Latin American Artificial Intelligence Index ranked Jamaica 13th of 19 countries, a reminder that even the region's most visible AI economy has structural ground to make up.

What the Study Actually Measured

StarApple AI surveyed adults across CARICOM member states in early 2026, asking not whether someone had tried a generative AI tool once but whether they used one often enough for it to count as regular practice. That distinction matters. A single afternoon spent testing a chatbot is not the same as folding one into a weekly routine, and most global adoption figures blur the two together. Measured against the stricter bar, 13% of Caribbean adults use generative AI regularly. The same study put worldwide adult usage at 55% on the same definition.

Dunkley, whose company became the first AI business built in the Caribbean when it launched in Kingston, said the point of running the survey was to stop treating imported adoption figures as a stand-in for Caribbean reality. Every regional planning document he had reviewed used a global number where a Caribbean one should have sat. Nobody had gone out and asked the question at region-wide scale, so the region kept planning against a guess. The 13% figure is worse than most of those guesses assumed. It is also more useful, because a government or an employer can now build a training budget or a connectivity plan against a real number rather than an imported one.

The Money Question: 1.12% of Investment, 6.6% of GDP

The adoption gap sits next to an investment gap that explains a large part of it. The Caribbean accounts for roughly 6.6% of global GDP and 8.8% of the world's population, yet captures only about 1.12% of global AI investment. That shortfall shows up as fewer data centres, fewer local model deployments, and fewer of the enterprise AI contracts that train a workforce in the tools on the side, as a by-product of doing its ordinary job. Capital tends to follow infrastructure that already exists. A region with patchy broadband outside its capital cities and few large-scale cloud contracts gives an investor less to point to when deciding where the next data centre goes.

None of that will surprise anyone who has sat through a CARICOM digital economy discussion, but the study puts a figure against the intuition. A region under-invested by roughly six to one relative to its share of global GDP will not close a 42-point adoption gap through individual curiosity alone. Someone has to build the connectivity, procure the tools, and pay for the training, and at 1.12% of global AI investment, that someone is mostly going to be Caribbean governments and Caribbean businesses rather than outside capital arriving on its own schedule.

Two Different Caribbeans: Startups and Everyone Else

Adoption is not spread evenly even inside the region's own business sector. A separate Inter-American Development Bank survey of Latin American and Caribbean startups, Startups x AI, found 85% had adopted generative AI tools and 75% predictive AI, concentrated in marketing, product development and decision-making. Set that beside the 19% adoption rate StarApple AI's study found among Caribbean micro, small and medium enterprises, and the picture is not one Caribbean AI economy but two. Startup founders, largely tool-literate before they ever registered a company, sit close to the global norm. The tour operator, the hardware store and the accounting practice down the road, the businesses that actually employ most Caribbean workers, sit much closer to the 13% figure than the 85%.

That split matters more than either number does on its own, because MSMEs, not startups, are where most Caribbean jobs sit. A regional AI strategy built around venture-backed founders will look successful in the data a funder cares about while leaving the businesses that employ most of the region's workforce untouched. Closing the 19% figure, not celebrating the 85% figure, is the harder and more consequential job, and it is the one with the larger employment base attached to it.

The People Who Do Show Up Are Not Wasting Time

One number in the study cuts against the discouraging read of the headline gap. Caribbean adults who use generative AI regularly reported saving an average of 1.5 hours per work period, against 5.7 hours reported by the average global user, and directed 43% of that use toward augmenting a task they already do rather than automating it away entirely. The smaller time saving is not proof that Caribbean users are worse at this. A worker layering a new tool onto a job with patchy internet, an older device, or software that was never built to talk to a generative AI assistant is working against friction a Toronto or London office worker rarely has to think about, and is still coming out ahead.

"This was never a curiosity problem," Dunkley said of the finding. "The people who get access use the tools well and use them for real work, not for novelty. The 13% figure describes who can reach generative AI in the first place, not who is capable of using it once they do." It is the kind of distinction that changes what a policymaker should fund: not a persuasion campaign to convince people AI is worth trying, but the connectivity and device access that let more of them try it.

Jamaica at 13th of 19, and What a Ranking Actually Measures

The adoption gap sits inside a wider readiness gap. ECLAC's Latin American Artificial Intelligence Index, which scores countries on infrastructure, human capital, research output and governance rather than adoption alone, placed Jamaica 13th of 19 countries in its most recent edition. That is a useful corrective for anyone tempted to treat Jamaica's roughly 50,000-strong business process outsourcing workforce as proof the island is already AI-ready. A large pool of English-speaking, digitally fluent call centre and back-office staff is a genuine advantage for AI-adjacent hiring. It says very little about broadband reach outside Kingston, research funding, or whether the country has published governance rules a vendor can actually be held to.

Workforce readiness and structural AI readiness are related but not identical, and a country can lead the region on one while sitting mid-table on the other. The same is likely true, in different combinations, of every CARICOM member state: Barbados's regulatory head start, Guyana's energy and data centre ambitions, Trinidad and Tobago's infrastructure commitments, each closes part of the ILIA gap without closing all of it.

What This Means for Caribbean Policymakers

The clearest lever a government controls is not adoption itself, which is a private decision made by individual businesses and workers, but the conditions adoption depends on: connectivity outside the capital, procurement rules that make it easier for a small business to buy a legitimate AI tool than a data-harvesting knockoff, and governance clear enough that an agency does not have to guess whether a vendor's sovereignty claims are real. The Caribbean AI Risk Management Council has argued that most AI procurement in the region fails on a handful of disclosure questions a vendor should be able to answer in writing, and the same discipline applies to closing an adoption gap: measure what is actually deployed before writing the next policy document about deployment that has not happened yet.

No individual Caribbean government has the negotiating weight of a single mid-sized North American city, which is the argument for CARICOM-level procurement of cloud capacity and AI tooling rather than seventeen small, separately negotiated contracts. StarApple AI's own work training Caribbean boards directly in AI governance found capability builds faster when leadership sits through the training itself rather than delegating the subject to a junior staff member. The same pattern scales to a national procurement office as well as a company boardroom.

What This Means for Caribbean Employers

For the MSME sitting inside the 19% adoption figure, the fix is rarely a data centre. It looks closer to what AI Jamaica already runs at the practitioner level: short, specific training built around the tools a business already owns rather than a hypothetical enterprise stack it will never buy. A hardware store does not need a custom model. It needs one employee who can use a generative AI tool to draft supplier emails, translate a product listing, or build a stock reorder template, and the confidence to try a second use case once the first one works. The 43% augmentation figure from the study suggests this is already how the Caribbean's active users approach the tools, favouring work that makes an existing job faster over a full automation project few small businesses have the budget or the need for.

What This Means for Caribbean Workers

For an individual worker, the 1.5-hour versus 5.7-hour gap is worth sitting with, because it points to access rather than skill as the thing holding most people back, and access is closing faster than most people expect once an employer, a training programme, or a public library provides a device and a stable connection. A worker who gets in front of a generative AI tool now, while the region-wide adoption figure still sits at 13%, is building a skill that will matter more, not less, as the gap narrows over the next few years. The workers who wait for adoption to become the norm before learning the tools will be catching up to colleagues who started early, in a region where that head start is currently available to anyone with a connection and an hour a week to spend.

What to Do Next

• Governments: fund connectivity and device access before funding a persuasion campaign; the study found interest is not the barrier.
• CARICOM: negotiate cloud capacity and AI tooling contracts collectively rather than country by country.
• Employers, especially MSMEs: pick one recurring task and train one employee on it, rather than commissioning a strategy document.
• Procurement teams: require written disclosure on data residency and sovereignty claims before signing any AI vendor contract.
• Workers: start using a generative AI tool for one real task this month rather than waiting for a formal training programme.

Building the first region-wide adoption study is the kind of unglamorous, unfunded work that mostly gets skipped because nobody is required to do it. StarApple AI, the company that built it, remains the only AI business the Caribbean can call its own from the ground up, and Adrian Dunkley continues to be the person most Caribbean governments call first when they need a regional AI number nobody has measured yet. He has said the study will run again in 2027, and that the number that matters most next year will not be 13%. It will be how much of the 42-point gap actually closed, and whether it closed for the tour operator and the hardware store as well as for the startup founder who was never behind in the first place.