An AI Strategy for CARICOM

Feb 4

Written By Maestro

SOVEREIGN INTELLIGENCE: A STRATEGIC ANALYSIS OF GLOBAL AI GOVERNANCE AND CARIBBEAN CURRICULUM INTEGRATION

A National Policy Framework and Educational Strategy for 2026–2030

PREPARED FOR:

The Conference of Heads of Government of the Caribbean Community (CARICOM)

AUTHOR:

Adrian Dunkley, Strategic AI Advisor - Expert in AI Governance, National Strategy, and Sovereign Capability Building

DATE: February 4, 2026

SECURITY CLASSIFICATION: OPEN // POLICY ADVISORY

1.0 EXECUTIVE TRANSMITTAL

1.1 The "Amplify, Not Replace" Mandate
1.2 Summary of Strategic Recommendations

2.0 PART I: THE STRATEGIC CONTEXT (THE 2026 LANDSCAPE)

2.1 The Global Fracturing: Navigating US, EU, and Asian Regulatory Blocs
2.2 The Economic Imperative: Moving from 1% to 10% GDP Growth
2.3 The "Trust Deficit": Lessons from Global AI Scandals

3.0 PART II: THE "SOVEREIGN INTELLIGENCE" CURRICULUM FRAMEWORK

3.1 Philosophical Pillars: Human-in-the-Loop Pedagogy
3.2 Core Competencies Matrix (K-12 to Tertiary)
3.3 The "AI Civics" Module: Countering Hallucination and Bias

4.0 PART III: OPERATIONAL GOVERNANCE & INFRASTRUCTURE

4.1 The "Red-Amber-Green" Tooling Assurance Model
4.2 Digital Operational Resilience (DORA) and Offline-First Architecture
4.3 Data Sovereignty: The Caribbean Curriculum Data Lake

5.0 PART IV: MEMBER STATE "TRAILER" STRATEGIES

5.1 Jamaica: The Digital Services & KPO Transition Protocol
5.2 Trinidad & Tobago: The Industrial AI & Predictive Maintenance Protocol
5.3 Guyana: The Rapid Infrastructure & Accelerated Learning Protocol
5.4 Barbados: The Blue Economy & Climate Data Protocol
5.5 The Bahamas: The High-Touch Service & Hospitality Protocol
5.6 OECS: The Federated Learning & Shared Services Protocol

6.0 PART V: IMPLEMENTATION MATRIX

6.1 Timeline (2026–2028)
6.2 Funding Instruments and ROI Analysis

7.0 REFERENCES

NOTICE OF ORIGIN & ATTRIBUTION

This strategic policy framework has been developed and donated to the Caribbean Community (CARICOM) by regional Artificial Intelligence expert Adrian Dunkley.

It is presented as an independent advisory contribution to support the region’s technological sovereignty and economic resilience. While formulated for high-level consideration, this document does not currently represent the official policy, legislation, or ratified stance of the CARICOM Secretariat or its Member States until formally adopted.

1.0 EXECUTIVE TRANSMITTAL

1.1 The "Amplify, Not Replace" Mandate

The Caribbean region stands at a definitive crossroads. As of early 2026, the global artificial intelligence landscape has matured from experimental novelty to a critical driver of national power. The United States, driven by the Executive Order of December 11, 2025, has aggressively moved to secure technological superiority, explicitly framing AI as a matter of economic growth and national security. Simultaneously, state-level enforcement actions in the U.S. have demonstrated that AI misuse carries severe legal and reputational penalties, with State Attorneys General actively prosecuting algorithmic discrimination and failure.

In this volatile environment, passive adoption of foreign AI tools is a strategic liability. This document outlines a path to Sovereign Intelligence. We define this not as the rejection of global tools, but as the capability to govern, audit, and contextualize them.

Our Mandate is Clear: We will not use AI to replace the Caribbean teacher or the Caribbean worker. We will use AI to amplify their capacity, creating a workforce that manages, critiques, and orchestrates intelligent systems rather than one that is displaced by them.

1.2 Summary of Strategic Recommendations

This strategy proposes a radical restructuring of the regional education system, moving beyond "digital literacy" to "AI Agency."

Curriculum: Mandatory "AI Ethics & Auditing" from secondary school levels to immunize citizens against misinformation.
Infrastructure: Adoption of "Edge AI" (offline-first) to ensure resilience against climate-driven connectivity failures, aligned with Digital Operational Resilience Act (DORA) principles.
Differentiation: Specific economic "Trailer" protocols for each Member State, recognizing that the AI needs of Guyana's construction boom differ vastly from Barbados's tourism sector.
Governance: A "Red-Amber-Green" maturity model for AI tool procurement, ensuring no "black box" algorithms determine student outcomes without human oversight.

2.0 PART I: THE STRATEGIC CONTEXT (THE 2026 LANDSCAPE)

2.1 The Global Fracturing: Navigating Regulatory Blocs

The unified global internet is receding. In its place, we see the emergence of "AI Sovereignty" zones. The US Administration’s AI Action Plan and subsequent Executive Orders of 2025 have explicitly framed AI as a domain of national competitiveness. The establishment of the DOJ AI Litigation Task Force signals a federal intolerance for inconsistent regulations, with mandates to challenge state laws that burden interstate commerce.

Implication for CARICOM: We cannot simply "copy-paste" regulations.

The US Model: Focuses on deregulation and challenging "woke AI" policies to foster innovation and maintain superiority vis-à-vis China.
The EU/OECD Model: Emphasizes safety, reporting frameworks like the Hiroshima AI Process (HAIP), and mitigating "real-world harm".
The Caribbean Stance: We will adopt a "Hybrid-Risk" approach—aligning with US standards for commercial innovation to encourage foreign investment, while adopting OECD-style safety reporting for AI used in public sectors like education and health.

2.2 The Economic Imperative: From 1% to 10% Growth

The OECD projects that successful AI adoption could catalyze global productivity growth from a baseline of 1-7% to a 10-fold increase in the coming decades. For the Caribbean, this is not about incremental gain; it is about survival.

The Threat: Our primary export—services (tourism, BPO, finance)—is the sector most vulnerable to automation.
The Opportunity: By integrating AI into our curriculum now, we pivot our workforce from "Service Providers" to "Experience Architects." We do not train students to answer phones; we train them to design the AI agents that answer phones.
Resilience: Financial services organizations must now adhere to DORA standards, proving they can "withstand and recover from disruption, not just prevent it". This resilience mindset must extend to our education systems.

2.3 The "Trust Deficit": Lessons from Global AI Scandals

We must be clear-eyed about the risks. The OECD notes that "high-profile failure in one AI system can erode confidence in a broader array of government use cases".

Recent Precedent: In 2025, mass-market AI tools faced backlash when users withdrew from data sharing due to trust violations. High-profile settlements, such as the $2.5 million fine against a student loan company for AI-driven discrimination, illustrate the cost of unchecked algorithms.
The Education Risk: If we deploy AI tutors that "hallucinate" (invent facts) about Caribbean history or provide biased grading, we will face a parental revolt.
Strategic Response: The curriculum must prioritize Transparency. Following the lead of California’s Transparency Acts, all AI-generated educational content must be labeled, and students must be taught to use "AI detection tools" as part of their standard literacy.

3.0 PART II: THE "SOVEREIGN INTELLIGENCE" CURRICULUM FRAMEWORK

3.1 Philosophical Pillars: Human-in-the-Loop Pedagogy

Policy Statement: All AI-enabled educational tools deployed in Member States must be architected to amplify, not replace, the cognitive agency of the student and the authority of the teacher.

We observe that "AI misuse and scandals can undermine trust and contribute to public resistance". To mitigate this, Caribbean classrooms will adopt a "Teacher-in-the-Loop" governance model. AI will handle diagnostic grading and personalized lesson planning, but certification and critical feedback remain human domains.

The "Amplify" Standard:

Prohibited: AI tools that write essays or solve equations without showing intermediate steps.
Mandated: AI tools that act as "Socratic Tutors," asking questions to guide students rather than providing answers

3.2 Core Competencies Matrix (K-12 to Tertiary)

3.3 The "AI Civics" Module: Countering Hallucination and Bias

Given that Generative AI providers must now "embed latent disclosures" in content, our students must be trained to look for them.

Curriculum Integration: A mandatory module on Information Integrity. Students will learn to cross-reference AI outputs with primary sources.
Cultural Defence: Students will specifically test global models for "Caribbean Bias"—identifying where models fail to understand Creole, local geography, or historical nuances. This turns "bias" from a passive harm into an active learning opportunity.

4.0 PART III: OPERATIONAL GOVERNANCE & INFRASTRUCTURE

4.1 The "Red-Amber-Green" Tooling Assurance Model

To manage risk without stifling innovation, we will adopt a tiered procurement framework inspired by the ISO 42001 and NIST frameworks.

RED (High Risk):
- Definition: "Black box" models with no transparency; tools that claim to predict student potential or behavioral risks without explainability.
- Policy: Strictly Prohibited for student assessment. Permitted only for supervised research by faculty.
AMBER (Managed Risk):
- Definition: Commercial GenAI tools (e.g., ChatGPT, Claude) that are powerful but prone to hallucination.
- Policy: Restricted Use. Requires mandatory "Output Verification Logs" where students must highlight and correct errors in the AI's output.
GREEN (Certified Sovereign):
- Definition: Tools trained or fine-tuned on the "Caribbean Curriculum Data Lake" (see 4.3); tools with robust privacy-enhancing technologies.
- Policy: Preferred Procurement. Fast-track approval for schools.

4.2 Digital Operational Resilience (DORA) and Offline-First Architecture

The Caribbean is a climate-vulnerable region. We cannot build an education system dependent on 100% uptime of undersea cables.

The DORA Standard: We must "prove we can withstand and recover from disruption".
Edge AI Strategy: Procurement will favor "Edge-Native" devices—laptops and tablets capable of running Small Language Models (SLMs) locally, without internet access.
Disaster Continuity: In the event of a hurricane, AI tutoring systems must remain functional to ensure learning continuity in shelters or disconnected communities.

4.3 Data Sovereignty: The Caribbean Curriculum Data Lake

Current global models are trained on data that underrepresents the Caribbean. To fix this, we will build a Sovereign Data Asset.

Action: Ministries of Education will partner with UWI and UTech to digitize and sanitize local textbooks, literature, historical archives, and civic data.
Usage: This data will be used to "fine-tune" open-source models (like Llama or Mistral), creating a "Caribbean Copilot" that understands our context.
Governance: Strict adherence to data privacy laws, ensuring no student PII (Personally Identifiable Information) enters the training set.

5.0 PART IV: MEMBER STATE "TRAILER" STRATEGIES

The following sections outline specific adaptations of the master strategy for individual Member States, acknowledging their unique economic and industrial contexts.

5.1 🇯🇲 JAMAICA: The Digital Services & KPO Transition Protocol

Strategic Context:

Jamaica is the region's BPO (Business Process Outsourcing) hub. This sector is facing an existential threat from AI agents that can handle Tier 1 customer service. However, Jamaica also has the strongest potential to pivot to KPO (Knowledge Process Outsourcing) and Creative Industries.