The AI Apocalypse Is Not What You Think
A Caribbean company can become more productive while the labour market becomes less welcoming. A Caribbean entrepreneur can gain global reach while the region loses control of the technology on which that success depends.
Artificial intelligence can distribute new capabilities to small firms and individual creators. It can also concentrate infrastructure, pricing power and accumulated knowledge in a handful of suppliers elsewhere. The region needs to manage both outcomes at once.
Three changes can happen without an alarm being sounded.
Lost positions are a small part of the story.
Employers may freeze entry-level hiring, reduce contracted hours and use fewer junior workers to produce the same output.
The first casualty is a job that never gets advertised.
Fewer routine roles can mean fewer desks and fewer chances to acquire the practical knowledge that produces future specialists.
One disruptive firm can thrive while an entire business district loses its centre of gravity.
Firms use imported tools, contribute data and recurring revenue, then depend on another country's suppliers for access, terms and continuity.
Lost positions are a small part of the story.
The Caribbean AI Risk Management Council (CAIRMC) has examined how exposure to AI-enabled tasks could translate into pressure on employment. Its Could AI Replace Me? five-year model covers 12 Caribbean economies with comparable inputs. In the central Jamaica scenario, gross displacement is equivalent to 5.3% of current positions. Separately, new demand could support work equivalent to approximately 2.5% of current employment. The regional gross displacement range is 3.1% to 6.0%. [1]
These are conditional outputs from a research model, not confirmed redundancies or probabilities of personal job loss. A role's tasks may change without its worker losing employment. The model makes assumptions about digital access, adoption and employers' decisions; it has not been independently peer reviewed.
New work does not automatically replace work that disappears.
Share of current Jamaican employment, five-year central scenario. Bar lengths share a common 0–6% scale.
Wider research supports examining tasks before counting jobs. The World Bank and International Labour Organization estimate that 26–38% of jobs across Latin America and the Caribbean have some exposure to generative AI; only 2–5% were classified as at risk of full automation under the study's assumptions. These are regional estimates, not Jamaica-only employment forecasts. [2]
A 2025 International Labour Organization study also concludes that job transformation is more likely than wholesale elimination for many occupations. [3]
Jamaica's related economic-scenario study highlights the difference between more output and better outcomes for workers. In a broad-adoption case, simulated GDP by January 2030 is 3.6% above the matched path without AI adoption. In a rapid-automation case, output is 13.5% higher, but unemployment is about 8.1%, compared with about 3.8% in the broad-adoption case. These are separate, assumed futures, rather than probabilities or an official outlook. [4]
Higher output can coexist with a weaker labour-market outcome.
January 2030 scenarios. GDP figures are level differences against matched no-AI paths. Unemployment is the modelled rate within each scenario.
Wider deployment, with some movement between jobs and new work.
Faster substitution, weaker movement between roles and fewer new cognitive tasks.
The first casualty is a job that never gets advertised.
Picture a graduate in Spanish Town searching for a first position in Kingston. A bank used to hire junior staff to reconcile accounts, prepare basic reports and answer standard requests. A services company used to recruit trainees to respond to routine customer queries. These jobs often gave people their first supervised exposure to exceptions, clients and decisions.
Now consider a firm that can handle much of that preliminary work with software. It retains senior staff, asks them to review the outputs and leaves two entry-level vacancies unfilled. No one receives a redundancy letter. Payroll may fall gradually through attrition, but the graduate loses a route into the profession.
Scale this mechanism across finance, insurance, customer support, legal administration and public services. Jamaica could face a shortage of experienced workers later because businesses stopped training beginners today. This is a plausible long-term risk; it is not an observed nationwide trend proven by the CAIRMC model.
The implication extends beyond salaries. Apprenticeship happens through doing. When junior employees no longer see routine cases, discuss mistakes and receive corrective feedback, the system may gradually lose the human judgement on which its own reliability depends.
Track graduate vacancies, junior hiring rates, contractor hours, wage progression, supervised training time and the share of routine tasks still performed by people. Headcount alone will miss the early change.
One new firm can thrive while an entire business district loses its centre of gravity.
There is real upside. A two-person team in Mandeville can use AI-assisted research, translation, design and customer support to pursue clients that once required a much larger payroll. A manufacturer in St Catherine may create new catalogues and service documentation faster. A tourism operator in Montego Bay can serve travellers in more languages without building a large back office.
For small businesses, AI can reduce the cost of testing ideas. That matters in a region where access to capital and specialised talent often limits growth. The opportunity is especially strong when local firms retain their customers, customer relationships, intellectual property and decision rights.
But the gains can be divided. Large employers may consolidate teams or shift routine services to foreign systems. A call centre with fewer agents may need fewer seats. A professional firm that stops recruiting juniors may lease less space. Lower office occupancy could then affect transport operators, food vendors, cleaning contractors and nearby retailers.
No cited index provides a measured forecast for Jamaican office-space reductions. This is a transmission pathway that commercial-property and labour researchers should test.
The physical reduction may be modest or delayed even where labour demand changes. Firms can repurpose space for higher-value work, expand into new services or retain staff for quality control. Property leases also adjust more slowly than staffing decisions. Good analysis should compare actual occupancy, hiring and local spending rather than assume every hour automated becomes an empty desk.
How an AI cost saving could reach the street outside the office.
How changes in staffing can affect surrounding commerce.
That is how the same technology could expand entrepreneurial opportunity and weaken an older model of shared prosperity. One founder wins overseas clients with a smaller team. A large operation stops hiring. The freelancer and the graduate compete for a narrower set of paid learning opportunities.
The impact depends on whether new firms replace lost demand, whether their income stays local and whether workers can move between sectors.
“Control shifts outwards. We feed the machine and then pay to use it.”Adrian Dunkley · Lead researcher, CAIRMC employment-risk research
We feed the machine, then pay for access to what it knows.
Jamaica and its neighbours have long relied on imported technology. What changes with AI is how much judgement, workflow and institutional memory may sit inside those tools. An overseas platform can become the place where a Caribbean company drafts reports, analyses customers, trains staff and makes recommendations.
The company may own its original data and decisions, yet still rely on the provider's software, compute capacity, policies, interface and pricing to get work done.
Data handling differs by provider, plan and contract; sending information to an AI service does not necessarily mean it becomes training data. The concern is structural dependency. A service can change its terms, rates, permitted uses, model behaviour or availability. A local firm that has not kept alternatives, documentation or internal expertise has little negotiating power.
Long-term reliance raises a second question: what happens to human capability? Research presented at CHI 2025 surveyed 319 knowledge workers and found an association between higher confidence in generative AI and lower self-reported critical-thinking effort during AI-supported tasks.
This was observational, self-reported research. It did not prove that AI causes permanent cognitive decline. The risk becomes more serious when organisations stop giving employees the chance to practise judgement independently. [5]
CAIRMC's separate Caribbean AI Risk Index offers a useful cross-check. Its released Jamaica profile records 64 points for labour-market exposure and 60 for cyber and AI-agent threats, alongside lower scores for its governance/readiness and resilience-gap pillars. These are comparative index points, not percentages of jobs at risk or probabilities of harm. The index is a descriptive framework rather than a validated forecast. [6]
Jamaica's risk profile points to specific pressure channels.
Five index pillars. Higher scores mean more exposure or a larger gap within each pillar's design. Shared 0–100 index scale.
Three risks can accumulate: economic leakage, as growing subscription and infrastructure costs flow outward; operational lock-in, when essential processes depend on one external service; and institutional forgetting, when people no longer understand the work well enough to verify it or continue without the tool.
The relative importance of each needs firm-level measurement. The concern is the direction of control, rather than a claim that international services have no value.
The region can gain efficiency and retain sovereignty.
A practical response begins with business choices, rather than slogans about replacing every imported model. Jamaica can use excellent external services while building bargaining power, retaining local expertise and developing products that serve Caribbean customers. Regional collaboration makes sense where no single economy has the scale to fund the full infrastructure alone.
- 01 Measure opportunity alongside employment changes. Publish changes in junior vacancies, internship places, contractor hours, wage growth and movement into skilled work alongside the employment rate.
- 02 Protect the learning pathway. Require paid apprenticeships, supervised case review and clear competence tests when employers automate entry-level tasks. Promote workers who can challenge a system's output.
- 03 Track real estate and surrounding businesses. Combine office occupancy, seat utilisation and lease renewals with transport, food-service and local sales data. Treat reductions as hypotheses to verify.
- 04 Buy AI with an exit plan. Require data-export rights, portability, security controls, disaster-recovery arrangements, vendor concentration limits and a way to keep essential services running during outages.
- 05 Own more of the value chain. Support Caribbean evaluation datasets, local language capabilities, domain-specific tools, trusted infrastructure, regional procurement and businesses that retain intellectual property and customer relationships.
| Risk to watch | Early signal | Accountable decision |
|---|---|---|
| Missing first jobs | Junior vacancies, training hours, starting salaries | Employers and education bodies preserve paid learning routes |
| Shrinking workplace footprint | Office occupancy, call-centre seats, lease renewals | Local planning bodies assess spillovers and reuse options |
| Weakening judgement | Independent error detection, unaided competency checks | Boards test human capability alongside tool adoption |
| Foreign-platform dependency | Vendor concentration, data portability, price exposure | Procurement teams maintain alternatives and exit provisions |
| Uneven gains | SME revenue growth, local contracts, wages and career access | Economic agencies assess who captures the added value |
The relevant question for the next decade is whether Caribbean institutions can produce more and develop more capable people at the same time. Productivity without opportunity can leave households behind. Innovation without ownership can leave firms dependent. A smaller company that wins a new export customer is a success; a stronger regional economy requires those gains to generate skills, income, competition and control that remain here.
Frequently asked questions
Will AI cause mass unemployment in Jamaica?
No reliable evidence establishes that outcome. CAIRMC's five-year model estimates gross position displacement of 5.3% and potential new-demand work equivalent to about 2.5% of current employment under its central Jamaica scenario. Those are assumption-dependent outputs, not recorded layoffs or a national unemployment forecast.
Why might artificial intelligence reduce office space in Kingston or Montego Bay?
Consolidating routine work, leaving vacancies unfilled and running smaller service teams may reduce demand for workstations, call-centre seats and leased office floors. The size and direction of any property-market effect remain unmeasured in the cited index. Expanding firms could offset some of that reduction.
What does CAIRMC's job-loss index measure?
It is a conditional employment scenario model. It uses broad employment shares, representative task profiles, access indicators and assumptions about adoption and employer decisions to estimate potential position displacement and new work. It does not observe individual redundancy decisions or predict the fate of a named worker.
What is meant by Caribbean AI sovereignty?
It means retaining meaningful control over sensitive data, critical business decisions, technical expertise, procurement terms and local value creation. It does not require every Caribbean organisation to build its own foundation model.
Can small Caribbean businesses benefit from AI without becoming dependent on it?
Yes. They can use AI to test products, serve customers and improve delivery while keeping records portable, retaining human review, checking provider terms and building staff who understand how the work is done without the tool.
Sources
- Caribbean AI Risk Management Council (2026). Could AI take your job? New Caribbean index puts 12 economies to the test. CAIRMC five-year employment scenario. Jamaica central case: 5.3% gross position displacement and approximately 2.5% new-demand work equivalents. CAIRMC Employment Index . Methodology .
- World Bank and International Labour Organization (2024). Generative AI and Jobs in Latin America and the Caribbean: Is the Digital Divide a Buffer or Bottleneck? Reports 26–38% employment exposure, 8–14% potential productivity gains and 2–5% risk of full automation in the wider region.
- International Labour Organization (2025). Generative AI and Jobs: A Refined Global Index of Occupational Exposure. Working Paper 140.
- Caribbean AI Association (2026). Jamaica: Jobs, AI and Export Demand. Conditional scenarios comparing GDP, wages and employment under different adoption and export-demand assumptions.
- Lee, H.-P., Sarkar, A., et al. (2025). The Impact of Generative AI on Critical Thinking: Self-Reported Reductions in Cognitive Effort and Confidence Effects From a Survey of Knowledge Workers. CHI 2025. Survey of 319 workers and 936 examples. Observational evidence rather than proof of long-term cognitive decline.
- Dunkley, A. / Caribbean AI Risk Management Council (2026). The Caribbean AI Risk Index — Method and Findings. CARI v2.1 research manuscript. Jamaica index pillars: labour exposure 64; cyber and AI-agent threats 60; information integrity 35; governance/readiness gap 31; economic resilience gap 30.
