At today’s GBA Summit, Vice President and former President, H.E. Dr. Bharrat Jagdeo, towards the end of his presentation, placed notable emphasis on the use of artificial intelligence to support technical work and analysis. It was not the central theme of his address. This author contends, however, that it may be one of the more consequential points made, because it speaks directly to the binding constraint Guyana now faces.
𝗟𝗮𝘀𝘁 𝘄𝗲𝗲𝗸, 𝗼𝗻 𝗧𝗵𝘂𝗿𝘀𝗱𝗮𝘆 𝟮𝟮 𝗢𝗰𝘁𝗼𝗯𝗲𝗿, 𝗜 𝗽𝗮𝗿𝘁𝗶𝗰𝗶𝗽𝗮𝘁𝗲𝗱 𝗶𝗻 𝘁𝗵𝗲 𝗽𝗮𝗻𝗲𝗹 𝗼𝗻 “𝗔𝗿𝘁𝗶𝗳𝗶𝗰𝗶𝗮𝗹 𝗜𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲, 𝗧𝗿𝗮𝗱𝗲 𝗮𝗻𝗱 𝘁𝗵𝗲 𝗙𝘂𝘁𝘂𝗿𝗲 𝗖𝗼𝗺𝗽𝗲𝘁𝗶𝘁𝗶𝘃𝗲𝗻𝗲𝘀𝘀 𝗼𝗳 𝗦𝗺𝗮𝗹𝗹 𝗘𝗰𝗼𝗻𝗼𝗺𝗶𝗲𝘀” 𝗮𝘁 𝘁𝗵𝗲 𝗪𝗼𝗿𝗹𝗱 𝗧𝗿𝗮𝗱𝗲 𝗖𝗲𝗻𝘁𝗿𝗲 𝗚𝗲𝗼𝗿𝗴𝗲𝘁𝗼𝘄𝗻’𝘀 𝗜𝗻𝘁𝗲𝗿𝗻𝗮𝘁𝗶𝗼𝗻𝗮𝗹 𝗧𝗿𝗮𝗱𝗲 𝗮𝗻𝗱 𝗜𝗻𝘃𝗲𝘀𝘁𝗺𝗲𝗻𝘁 𝗖𝗼𝗻𝗳𝗲𝗿𝗲𝗻𝗰𝗲 (𝟭𝟵–𝟮𝟯 𝗢𝗰𝘁𝗼𝗯𝗲𝗿 𝟮𝟬𝟮𝟲). The question I put forward was this: how do we move from resource abundance to productivity abundance by leveraging AI? I argued that, for a small economy, AI should not be treated as a technology race, but as a productivity, institutional-capacity and competitiveness strategy.
The issue must be examined relative to where Guyana now stands. We are moving from managing financial scarcity to managing resource abundance. The binding constraints are no longer primarily financial; they are execution, specialised skills, data quality, coordination and time. Money becomes less scarce precisely when institutional capacity becomes more valuable. Productivity abundance, as I frame it, is the stage where that capacity, rather than the resource endowment alone, drives sustained non-oil growth.
The public sector shows this clearly. As I noted at the panel, public-sector employment rose from about 60,000 in 2020 to about 80,000 today, an addition of roughly 20,000 workers in six years. That expansion has taken place in an economy already short of both labour and skills, and with the scale of the development agenda still ahead, that shortage is not going away. Each new public-sector hire comes from the same limited pool that private firms, contractors and investors are drawing on. Hiring our way to capacity therefore has clear limits: it raises the recurrent wage bill and adds to wage pressure without, by itself, raising output per worker. The more sustainable route is to raise what each worker can deliver.
That is the case for AI. At the panel, I presented simulation results from our SphereX study, From Resource Abundance to Productive Abundance: The AI-powered Leapfrog Strategy for Guyana. Under the central scenario, AI-enabled workflow redesign could release about 10.1 million public-sector hours a year, equivalent to roughly 4,864 full-time staff (about a quarter of the headcount added since 2020), and about 48,600 FTE-years over a decade. The simulations put the annual service and capacity value at GYD 13.8 billion, rising to GYD 18.9 billion in annual public value by Year 5, with a five-year net present value of GYD 35.3 billion. On the macro side, non-oil output would sit about 1.8% above the no-AI baseline after five years and 3.6% after ten, equivalent to about GYD 120.8 billion in annual non-oil GDP uplift by Year 10. That implies a productivity-growth proxy of about 0.35 percentage points a year, and possible CPI relief centred near 0.4 percentage points. These are simulations based on stated assumptions, not forecasts or booked savings. Released time becomes economic value only when institutions redesign workflows, redeploy capacity productively and independently verify the gains.
Faster appraisal, procurement and permitting shorten the time between budget allocation and a productive asset. Shorter delivery cycles reduce cost overruns and idle capital. Earlier commissioning brings forward the income-earning potential of public investment and lifts the economy’s absorptive capacity without expanding the wage bill at the same rate.
The principal challenge, however, is not access to an AI model; it is whether our data, workflows and accountability arrangements are ready. If we automate a badly designed workflow, we simply produce inefficiency at greater speed. The validated 2021–2026 procurement data illustrates this: roughly 1% of suppliers captured 25% of contract value, and the top 10% captured about 70%. An AI layer that merely digitises that structure adds little. One that improves supplier discovery, bid compliance and milestone verification can widen the domestic delivery base.
Governance is equally non-negotiable. AI is assistive, not authoritative: AI proposes; authorised humans decide. And in a labour-short economy, the objective should be augmentation, not displacement: retraining and redeploying people to where they are most needed, rather than redundancy.
In light of the foregoing, the Vice President’s emphasis is well placed. For Guyana, the real leapfrog is not spending more; it is shortening the time between expenditure and productive outcomes. As I said in closing at the panel:
“𝗚𝘂𝘆𝗮𝗻𝗮’𝘀 𝗼𝗽𝗽𝗼𝗿𝘁𝘂𝗻𝗶𝘁𝘆 𝗶𝘀 𝘁𝗼 𝗰𝗼𝗻𝘃𝗲𝗿𝘁 𝗿𝗲𝘀𝗼𝘂𝗿𝗰𝗲 𝗮𝗯𝘂𝗻𝗱𝗮𝗻𝗰𝗲 𝗶𝗻𝘁𝗼 𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝘃𝗶𝘁𝘆 𝗮𝗯𝘂𝗻𝗱𝗮𝗻𝗰𝗲. Responsible AI can multiply scarce expertise, improve institutional throughput and lower the transaction costs of trade and investment. But the decisive variables will remain human capability, workflow redesign, trusted data, governance and disciplined implementation. AI can accelerate development; it cannot substitute for competent institutions.”