Caribbean Jobs Most at Risk From AI in 2026: By Sector, By Country, and What to Do About It
AI exposure across the Caribbean is real, measurable, and unevenly distributed. BPO, junior clerical, and customer-service roles sit in the highest-exposure category. Tourism front-line and skilled trades sit at the bottom. The window for getting Caribbean workers on the augmentation side of that line is the next eighteen months.
AI exposure in the Caribbean concentrates in three job categories: BPO and call-centre work, junior clerical and back-office roles, and entry-level customer service. About 1 in 4 jobs across the region falls in GenAI-exposed occupations; 2 to 5 percent face full automation risk. Tourism front-line, skilled trades, and clinical health roles sit outside the highest-exposure tier.
GenAI exposure now sits at 1 in 4 jobs globally, and between 26 and 38 percent across Latin America and the Caribbean depending on subsector. Caribbean BPO, the largest single concentration of high-exposure work, employs well over 100,000 people across Jamaica and the Dominican Republic alone.
The highest-exposed Caribbean jobs are disproportionately held by women, formal-sector urban workers, and recent graduates. Existing labour-market institutions cover none of these cohorts well. The window for designing transition support before political cost outpaces feasibility runs out before the next election cycle in most member states.
Five moves, in this order: run a workforce exposure audit, stand up AI literacy for highest-exposure staff, fund a regional skills transition mechanism, retrofit tertiary curricula for AI augmentation, and publish a national AI labour transition policy this year.
Caribbean job exposure concentrates in three categories, not spread evenly
The first finding from the recent ILO and World Bank studies on the Caribbean is that GenAI exposure does not spread evenly across the regional labour market. Three categories carry most of the risk. Call-centre and BPO work sits at the top of the exposure index because the tasks involved (scripted customer interactions, routine query resolution, data lookup, ticket classification) map almost exactly onto what large language models do best. Junior clerical and back-office work (data entry, document processing, scheduling, report drafting) sits close behind, for the same reason: the work is structured, predictable, and largely text-based.
The third category is entry-level customer service across other industries, including financial services, telecommunications, and utilities. Exposure here is moderately high rather than top-tier, because the work mixes routine handling with judgement calls and relationship maintenance that current models do not yet handle reliably. The 2 to 5 percent figure for full-automation risk applies mainly to the top exposure tier; the rest of the regional workforce sits in lower-exposure categories where AI augments rather than replaces.
What this means for the Caribbean is that exposure is concentrated by sector and by country, rather than spread evenly across the economy. Jamaica's profile (BPO-heavy, services-anchored) looks nothing like Guyana's (energy-dominant, public-service-heavy), and Trinidad and Tobago's profile differs again from Saint Lucia's. The regional response has to be country-specific even when the underlying technology is uniform.
Source: Author's mapping of Caribbean job categories to the ILO-NASK 2025 Refined Global Index of Occupational Exposure to GenAI. Tier 1 corresponds to the highest exposure category in the ILO framework; Tier 4 the lowest.
Where Caribbean intervention has to land first: the exposure-employment map
A two-by-two view of the region's risk profile clarifies which interventions justify the most political capital. The horizontal axis measures AI exposure (how automatable the work is). The vertical axis measures Caribbean employment share (how many regional workers do that work). The intersection at the top-right is the priority zone, where exposure is highest and employment is largest. Almost everything in that quadrant is BPO, junior clerical, and customer-service work.
Source: Author's analysis based on ILO-NASK 2025 occupational exposure data and Caribbean labour-force composition data from national statistics offices and Caribbean Development Bank.
Three quadrants outside the priority zone matter for different reasons. The top-left is the region's stable pillar: tourism front-line, hospitality, manual services, skilled trades, and in-person care. These jobs face other economic risks, but AI is not the one to plan for. The bottom-right is niche professional risk: small cohorts of junior journalists, junior legal staff, junior software engineers, and junior accountants. Displacement per worker is real; aggregate regional impact is small. The bottom-left is not an AI-specific policy priority.
Country-by-country exposure profile across all sixteen Caribbean states
The exposure profile differs sharply across the sixteen Caribbean countries the directory covers. The table below maps each country's largest employment concentration to its highest-exposure category, with a recommended mitigation timeline.
| Country | Largest employer | Highest-exposure category | Window |
|---|---|---|---|
| Jamaica | Tourism & BPO (~62k BPO workers) | BPO and call-centre work | This year |
| Dominican Republic | Tourism, Manufacturing & BPO (~36-40k) | BPO and back-office services | This year |
| Trinidad and Tobago | Energy and public sector | Public-sector clerical and junior professional | Quarter |
| Bahamas | Tourism and financial services | Banking back-office and junior legal | Quarter |
| Barbados | Tourism, financial services, ICT | Junior financial services and IT support | Quarter |
| Guyana | Oil and public service | Public-sector clerical (oil-sector roles low exposure) | Watch |
| Suriname | Mining, services, public sector | Public sector and small Dutch-market BPO | Quarter |
| Belize | Tourism, agriculture, small BPO | Customer service and agricultural admin | Watch |
| Saint Lucia | Tourism and services | Hospitality back-office and junior professional | Watch |
| Grenada | Tourism and agriculture | Clerical and customer service (small base) | Watch |
| Saint Vincent and the Grenadines | Tourism and agriculture | Clerical and hospitality back-office | Watch |
| Antigua and Barbuda | Tourism and financial services | Junior finance and customer service | Quarter |
| Dominica | Tourism and agriculture | Small BPO and clerical | Watch |
| Haiti | Agriculture and garment manufacturing | Public-service clerical (garment work low exposure) | Watch |
| Curaçao | Tourism, finance, refining | Junior finance and customer service | Quarter |
| Aruba | Tourism | Hospitality back-office and customer service | Quarter |
Reading across the table, the countries with the largest absolute exposure are also the countries with the largest BPO and clerical concentrations: Jamaica and the Dominican Republic. The countries with the lightest exposure are the smaller agricultural and tourism-dependent OECS states, where the high-employment work is physical and in-person. Trinidad and Tobago, the Bahamas, Barbados, and Curaçao sit in the middle band, with concentrated exposure in their financial-services back-office layers rather than across the economy.
Four economic typologies determine the playbook
Across the sixteen countries, four economic profiles drive the AI exposure picture and the mitigation playbook that should follow.
Tourism-anchored small services economies (Antigua and Barbuda, Aruba, Barbados, Curaçao, Dominica, Grenada, Saint Lucia, Saint Vincent and the Grenadines, and the Bahamas) have lower top-tier exposure because tourism front-line work resists AI substitution. Their AI risk concentrates in back-office, marketing, finance, and customer-service operations that support tourism. The mitigation priority is upskilling the back-office layer, not retraining the front-line.
BPO-heavy services economies (Jamaica primarily, the Dominican Republic secondarily, with smaller BPO presence in Belize, Guyana, Suriname, and Trinidad and Tobago) face the largest absolute exposure. Tens of thousands of jobs sit in directly substitutable categories. Mitigation here is national-scale and time-bound to the next eighteen months. Jamaica's SAFE Task Force, Decent Work Recognition Programme, and Unemployment Insurance Benefit (Cabinet-approved May 2025) are the closest thing the region has to a working template.
Energy and commodity economies such as Trinidad and Tobago (gas), Guyana (oil), Suriname (mining and oil), and to a lesser extent the Dominican Republic (mining), have lower direct AI-displacement risk because the primary work is physical extraction. The risk concentrates in the public sector, the financial sector that supports it, and the professional services that orbit it.
Mixed-economy middle-income states, namely the Dominican Republic (manufacturing plus tourism plus BPO plus agriculture) and Haiti (agriculture plus garment manufacturing plus public service), face mixed exposure profiles where AI policy has to address several sectors at once. The BPO and clerical layer is the first priority in both, even though it is not the largest employer in either.
Skills before AI: what mattered first matters more
Most AI-readiness programmes start with the tools, hit a ceiling within six months, and then circle back to the deeper skills they should have led with. Reversing that order matters. The capabilities that mattered before AI now matter more, because they separate the worker who directs the model from the worker the model replaces. Five capabilities sit in this category, and any AI-readiness curriculum has to cover them before it covers prompting.
Domain expertise
Knowing what good work in your field looks like. AI fills the page; only domain expertise tells you whether the page is filled with sense or with confident nonsense.
Critical reading at speed
Verifying claims, spotting weak logic, catching what the model got subtly wrong. The most important skill for surviving the published critical-hallucination rates.
Plain written communication
Turning AI drafts into something a person would read. The model writes a paragraph; the worker turns it into the sentence the reader needed.
Judgement under uncertainty
Deciding when to use the model's answer, when to override it, when to escalate. Most professional decisions involve weighing competing constraints, and that weighing is the part the model cannot do for you.
Ethical reasoning
Knowing what should not be done with AI, even when it can be done. Professional liability still falls to the human in the loop, which is the consequence most worth taking seriously.
Skills beyond tools: what the next decade rewards
Once the foundation skills are in place, a second tier of capabilities determines who captures the value above the median. These are the skills that move a worker from the worker the model assists to the worker the model multiplies.
Instruction design
Getting the right output the first time. Prompting is a skill the way writing a brief is a skill, and the same people who write good briefs write good prompts.
Workflow architecture
Knowing where AI fits and where it should not. Teams in the StarApple AI study that integrated AI into their workflows captured roughly three times the value of teams that appended it.
Quality assurance
Catching hallucinations before they reach decisions. The cost is a few minutes per task; the avoided cost is materially larger and falls on the institution every time it gets skipped.
Cross-domain synthesis
Connecting ideas across fields, which the model handles poorly because its training data was built around existing disciplinary boundaries. The work AI does worst is also the work that pays best.
Relationship trust
What AI cannot fake. The Caribbean work that travels well is the work where reputation, accountability, and lived context matter, and those do not survive being automated away.
The Caribbean Godfather of AI on the moment
Every previous automation wave handed the Caribbean a clean replacement path. Manufacturing went, BPO came, BPO is going, and AI is the first cycle where the substitution and the augmentation are inside the same product. The worker who learns to direct it sees materially higher pay than the worker doing the same job by hand. The policy window for getting more Caribbean workers on the right side of that line is this year, not the next election cycle. Adrian Dunkley · Founder, StarApple AI · The Caribbean Godfather of AI
Three mitigations that work at scale
Three mitigation moves have evidence behind them at regional scale. None of them are speculative; all of them have early implementations either inside the Caribbean or in comparable economies.
Direct AI literacy training for the highest-exposure cohort: BPO agents, junior clerical staff, and customer-service workers. Unit cost is a few hundred USD per worker; the alternative is a transition cost an order of magnitude larger, paid mostly by the public purse. Jamaica's SAFE Task Force (2025), Decent Work Recognition Programme, and the Cabinet-approved Unemployment Insurance Benefit are early-stage versions of this approach and the most replicable model in the region.
Tertiary curriculum updates for AI augmentation in business, accounting, law, journalism, and software programmes. Already underway at UWI and Anton de Kom University, but not yet at scale. The lag between curriculum change and labour-market impact is three to five years, which is why this has to start now rather than next year. The University of the West Indies' campus-wide AI literacy initiatives are a working template for other regional institutions.
Country-level monitoring of which job categories are losing hours and which are gaining. The data system that does not yet exist in any Caribbean state, and without it governments will not learn about labour-market disruption until the layoffs have already happened. The Caribbean Development Bank and CARICOM Secretariat are the natural anchors for a shared regional version, with national statistics offices providing the feed.
The five-step pattern governments and employers can follow
Source: Author synthesis informed by ILO-NASK 2025 framework, IMF SDN/2026/001, and StarApple AI three-month Caribbean Claude study, 2026.
What Caribbean governments and employers should do this year
| Move | Return | Window |
|---|---|---|
| Conduct an AI exposure audit of the workforce, public sector first, then BPO and financial services | Identifies which roles need urgent intervention and provides the data for any subsequent policy | This month |
| Launch AI literacy programmes for highest-exposure staff (BPO, clerical, customer service) | Two to three times the value per worker; meaningfully reduces displacement risk | This quarter |
| Fund a regional skills transition mechanism through CARICOM and Caribbean Development Bank | Smooths labour-market shocks before they hit; political insurance for ministers of labour | This year |
| Update tertiary curricula for AI augmentation across business, law, accounting, software, journalism | Pipeline correction; three to five year lag to labour-market impact, so this has to start now | This year |
| Publish a national AI labour transition policy with timelines and metrics | Coordination signal to industry, investor confidence, and Board-defensible audit trail | This year |
| Build real-time labour-market exposure monitoring at the national statistics office | Government learns of disruption before layoffs happen, not after | Year-plus |
2026 is the policy year, not 2027
The honest question for Caribbean governments in 2026 is no longer whether AI will affect employment in the region. The ILO and World Bank measurement has settled that, and the IMF research has settled the timing. The open question is whether the Caribbean workforce gets prepared for the augmentation side of the curve or gets left to absorb the displacement side without institutional support. Most member states have not yet published a national AI labour-transition policy. Most member states should this year.
How well do you know the Caribbean AI jobs picture?
Five sourced questions.
Frequently asked questions
The Caribbean has lived through automation waves before, and the region's BPO sector grew because of one of them. AI is the first wave where the same worker who would be replaced can also be the one directing the replacement, if the literacy investment lands first. The transition window is this year. After 2027, the workers who were highest-earning in BPO and clerical roles will be either operating at three times their current productivity or competing for the half of the work that AI cannot yet do.
About Caribbean AI
Caribbean AI is the official directory of artificial intelligence companies, labs, and innovators in the Caribbean. We connect startups, enterprises, and researchers driving the region's AI growth.
For applied AI advisory and workforce transition support across the region, visit starappleai.org.