Method · 31 October 2026
How the CAIRME index works
The map estimates separate five-year scenarios for positions displaced and positions supported by new demand. It combines a country’s broad employment mix with explicit assumptions about task substitution, adoption and employer response. The role tool lets you replace several assumptions with conditions at your workplace.
What comes from country data
Employment shares for agriculture, industry and services are the World Bank’s 2025 ILO modelled estimates. Individual internet use is the World Bank/ITU indicator for 2024, except Suriname (2023). The latter is an access proxy in the adoption scenario; it is not a count of firms using AI. Four CARI countries lack the three comparable employment shares, and remain unscored. Territories remain visible without a national percentage. Sector source · Access source.
How tasks become a scenario
For each task, the model multiplies time share, AI capability, feasibility and the portion that does not require human or physical work. Agriculture and industry use representative task profiles. Services use the simple mean of ten service-role profiles, not a measured occupational mix within each island. The map estimates adoption from a starting value of 0.12 + 0.18 × internet-use share, growing at 15% annually for five years in the current pathway. It then assumes that 45% of feasible task substitution translates into position reduction. A separate 30% demand-capture factor converts augmentation capacity into potential jobs.
Displacement: Σ (sector employment share × adoption × 0.45 × substitutable task share)
Potential creation: Σ (sector employment share × adoption × 0.30 × augmentable task share)
Slower and faster pathways multiply adoption by 0.75 and 1.25. The displayed range varies adoption growth, position-reduction response, creation capture and task coverage in 1,600 repeatable draws. This is a sensitivity analysis of chosen inputs.
What the results show
For the 12 economies with 2025 sector baselines, the current pathway spans 3.1% to 6.0% gross displacement and 1.7% to 2.7% potential creation as shares of current employment. The Bahamas has a larger services employment share than Haiti in the ILO series, which contributes to their different scenario values. These are outputs of the stated formula; they are not observed job losses or a ranking of individual workers.
What would improve the estimate
Country labour-force data at the occupation and industry level, local task-time studies, employer adoption surveys, workplace computer access, vacancies and longitudinal worker transitions would replace broad-sector assumptions. A prospective study should freeze the calculation before observing outcomes, track firms and workers over time, and test calibration against actual staffing and hiring. Anthropic’s observed-exposure method motivates a useful next input: distinguish tasks that AI can perform from tasks used in automated work settings. Anthropic study · ILO–World Bank regional study.
How to use it
Workers can map actual time spent on routine outputs and exceptions, then seek paid learning and redeployment discussions. Employers can measure task quality, demand and staffing effects in pilots before choosing a labour response. Policymakers can collect the missing data and monitor hiring as well as displacement. The detailed paper is held as a research draft for author review before preprint submission.