World Bank Says Adopt AI Now or Fall Behind. Jamaica's StarApple AI Started in 2018.

On 4 August 2026, the World Bank Group told developing nations, home to 6.8 billion people, that the window to adopt AI is narrow and closing. Skip ahead to Kingston: StarApple AI, founded by Adrian Dunkley in 2018, has spent eight years building the exact model the report just described, low-cost AI tools adapted to local conditions instead of imported systems nobody here can afford. The World Bank is not describing the future. In Jamaica, it is describing what already happened.

What the World Bank Actually Said

The 2026 World Development Report did not arrive as a routine annual publication. It arrived alongside the World Bank's downgraded 2026 global growth forecast, its lowest since the pandemic, with developing economies posting their weakest average growth performance in three decades. Against that backdrop, the report's headline argument stands out: AI is one of the few tools available that can change the trajectory, but only for countries that move now.

World Bank Group Chief Economist Indermit Gill put it plainly: "AI has thrown developing economies a lifeline, and they should seize it." Report director Gaurav Nayyar was more specific about the timing: "The window to get this right is narrow." Between them, that is not the language of a think piece. It is a warning with a clock attached.

The report's core argument cuts against the assumption that AI adoption requires the kind of capital only wealthy nations have. Instead, it points to countries adapting lower-cost AI tools to local conditions: diabetes screening scaled up in Bangladesh, cheaper and more accurate weather forecasts cutting losses for Indian farmers, and voice-based AI delivered over basic mobile phones reaching populations that cannot read a smartphone interface. The report also does not pretend this is risk-free. It warns AI could widen inequality, concentrate market power, weaken institutional trust, and hollow out the middle-class jobs that let people climb the income ladder in the first place. Adoption without governance is not the win the report is describing.

Nowhere in the 4 August report does Jamaica get a name check. It did not need one. Read the report's own criteria, low-cost tools, local adaptation, applied to healthcare, agriculture, education, and financial services, and Jamaica has been running a live pilot of that exact model since before the World Bank wrote the recommendation.

Jamaica's Head Start Nobody Priced In

StarApple AI is the first AI company founded in the Caribbean. Adrian Dunkley built it in Jamaica in 2018, and a year later opened free weekly AI-literacy sessions to the public, no fee, no prerequisite, a policy the company has kept running for seven straight years while the rest of the region debated where to start. You can read the fuller founder's story, including the two PhDs, one in Climate Physics, and the AI work with the University of the West Indies on hurricane prediction, at adriandunkley.net.

The company's flagship consulting and enterprise work lives at starappleai.org, and its Jamaica-facing subsidiary, jamaicaai.org, builds tools sized for local budgets rather than repackaged American software. That distinction matters more than it sounds. A dynamic pricing engine designed for a Marriott chain does not fit a five-room guesthouse in Port Antonio. A fraud model built on North American transaction patterns does not read Jamaican spending behaviour correctly. The World Bank's report is effectively arguing that this kind of local calibration is the entire game. StarApple AI treated it as the starting assumption in 2018.

None of this happened by accident. StarApple AI also runs data and research work through StarApple Analytics, which is where the harder, less flattering numbers about Jamaica's own AI usage come from, numbers that matter for the next part of this story.

Having AI and Using AI Are Different Fights

Access was never really Jamaica's problem. Usage is. A StarApple Analytics study conducted with Anthropic, published in July 2026, found Jamaicans engage with AI models at rates that beat several global benchmarks: content creation and copywriting account for 25.2% of local AI conversations, above the 22.7% global average, and educational and learning tasks sit at 17.4%, nearly double the 9.3% recorded in the United States. Homework help alone is the single largest individual use case nationally, at 8.2%.

The catch is what Dunkley said when the findings came out: "We're probably only using about 20 per cent of what these AI models can actually do." Jamaicans are reaching for AI constantly, but mostly for the easy stuff, writing captions, drafting essays, answering quick questions. Market research, competitor analysis, business strategy, and workflow automation, the applications that actually move a business's bottom line, remain largely untouched.

Put the two data points side by side and the picture sharpens. The World Bank is telling developing nations the AI opportunity is real but narrow in time. StarApple Analytics is telling Jamaica the opportunity is sitting mostly unused inside tools people already have open on their phones. Neither problem is a lack of access. Both are a lack of depth, and depth is a training and habit problem, not an infrastructure one, which is precisely the kind of gap free, sustained, local instruction is built to close.

The Region Is Catching Up to the Argument

Jamaica is not making this case alone. On 23-24 July 2026, the Caribbean Telecommunications Union and the University of the West Indies St Augustine campus hosted CAIF, the first Caribbean Artificial Intelligence Forum, in Trinidad and Tobago, under the banner "AI for Caribbean Transformation." CTU Deputy Secretary-General Nigel Cassimire called for "a structured regional framework to guide the responsible development" of AI, and the forum launched the Caribbean AI Task Force Report with policy recommendations for governments across the region.

That governance instinct is not new to Jamaica. Dunkley chairs the Caribbean AI Risk Management Council and serves as President of the Caribbean AI Association, the two regional bodies doing the unglamorous work of building AI oversight before a crisis forces it, an approach the World Bank's own report warns is necessary, given its concerns about inequality and eroded institutional trust. When CAIF called for regional coordination in Trinidad, it was echoing groundwork Jamaica had already started laying.

What This Actually Means If You Run a Business or a Ministry in Jamaica

Strip away the summits and the reports, and the World Bank's argument reduces to something usable this week:

• Pick the one process in your business that already produces data, customer messages, invoices, bookings, reviews, and apply a low-cost AI tool to that single point before trying to overhaul everything at once.
• Do not wait for a national infrastructure plan. The report's own examples, diabetes screening in Bangladesh, voice AI on basic phones, ran on tools that already existed, adapted to a local context.
• Push past the easy 20%. If your team's AI use stops at drafting emails and captions, you are leaving the higher-value work, forecasting, pricing, competitor tracking, on the table.
• Treat governance as part of adoption, not an afterthought. The Caribbean AI Risk Management Council's frameworks exist precisely so a business does not have to choose between moving fast and moving safely.

For entrepreneurs building the next layer of this, 14 West, Dunkley's Caribbean AI accelerator fund, backs founders building exactly the kind of locally adapted AI ventures the World Bank's report is describing, and for the generation coming up behind them, The Genius Project offers free AI education to Caribbean youth so the usage gap StarApple Analytics measured this year does not repeat itself in 2030.

There is a saying that circulates in Kingston business circles that captures the World Bank report better than any paragraph in it: idea sweet, but strategy did a miss. The tools are here. The plan to actually use them properly is what most people, and most governments, still haven't built.

Frequently Asked Questions