How to build a workforce scenario with AI

Frame the decision before the scenarios

State the time horizon, workforce in scope, business decisions the analysis must support, and factors outside scope. A scenario for near-term training allocation needs different detail from a five-year location strategy.

The International Labour Organization describes skills anticipation as a systematic process for identifying and preparing for future skills needs. Its foresight and scenario guide combines qualitative and quantitative approaches and emphasizes adapting methods to context. These sources support using multiple futures, not any specific organizational forecast.

Identify independent demand drivers

Choose a small set of uncertain forces that materially affect work. Examples include demand volume, task automation, regulation, service design, product mix, and labor supply. Avoid duplicate drivers such as “AI adoption” and “automation rate” unless their meanings differ.

For each driver record the current evidence, plausible low and high states, leading indicators, and owner. Do not ask AI to invent ranges from general market commentary.

Worked synthetic scenario set

Suppose an assessment operations team has 20 reviewers. Leaders need to decide whether to train current reviewers, recruit specialists, or redesign the workflow over 18 months. Two uncertain drivers dominate: assessment volume and the share of routine evidence-linking that approved tools can assist.

| Scenario | Volume | Assisted routine work | Workforce implication | | --- | --- | --- | --- | | Steady craft | Low growth | Limited | Maintain capacity; deepen complex-review skill | | Scaled review | High growth | Moderate | Train for exception handling and add reviewer capacity | | Redesigned flow | Moderate growth | High | Fewer routine steps; greater demand for verification and escalation | | Capacity squeeze | High growth | Low | Backlog risk; redesign intake and recruit selectively |

These scenarios do not predict which future will occur. They make different preparations discussable.

Translate workload into roles and skills

For each scenario, estimate task volume, frequency, handling time ranges, feasible assistance, quality constraints, and skills needed. Show which roles expand, contract, or change. A tool reducing one task does not establish that a role disappears; other tasks, controls, and demand may absorb the capacity.

Ask AI to check arithmetic, trace every role change to a driver, and find skills that recur across scenarios. Those common capabilities are candidates for no-regret investment.

Add trigger indicators

Attach observable signals that show a scenario is becoming more relevant. Examples include sustained assessment arrivals, verified tool-error rates, review rework, time-to-competence for new reviewers, and policy changes. Define the threshold, observation window, and decision it triggers.

For the synthetic team: if arrivals exceed demonstrated reviewer capacity for six consecutive weeks while rework stays within the quality boundary, capacity action begins. If assisted evidence-linking passes controlled quality checks, training shifts toward exception review. These are illustrative governance choices, not recommended universal thresholds.

Stress the plan

Change one assumption at a time: demand, productivity, attrition, learning time, or implementation delay. Then combine adverse assumptions. Keep ranges visible and distinguish headcount, full-time-equivalent capacity, skills, and budget.

Deliver a decision packet

Include driver cards, scenario narratives, workload model, role and skill implications, common actions, scenario-specific options, trigger dashboard, assumptions, and review schedule. State where evidence is weak.

Verification checklist

Confirm the scenarios are meaningfully different; drivers are evidence-linked; calculations reconcile; role changes trace to tasks; supply constraints include learning time; indicators are observable; decisions have owners; adverse combinations are tested; and the packet avoids individual employment recommendations.

Sources and claim map

The team, scenarios, quantities, and triggers are synthetic. This educational method does not forecast demand, validate staffing levels, or make automated employment decisions.

• International Labour Organization, Anticipating and matching skills and jobs: supports the systematic skills-anticipation framing and the use of data, methods, tools, and institutions.
• International Labour Organization, Developing skills foresights, scenarios and forecasts: supports combining quantitative and qualitative approaches for contextual scenario work.

Sources: [1] [2]