Practical guide
AI fluency for workforce planning: test task assumptions
Model a fictional workload change using task-level assumptions without converting an AI time-saving estimate into a headcount decision.
Workforce planning managers can practice AI fluency by testing task assumptions before drawing conclusions about capacity. This fictional scenario separates preparation time, review time and new checking work. It produces a sensitivity worksheet, not a staffing recommendation, an employment decision or evidence that AI delivers a measured productivity gain.
Break the work into components
A single time-saving percentage can hide work that remains or moves elsewhere.
The fictional team prepares ten weekly briefs. Each brief currently assumes forty minutes of drafting and twenty minutes of review, for sixty minutes total. These are scenario inputs, not measured workplace timings. Ten briefs therefore represent six hundred minutes under the stated model.
A proposal says AI cuts drafting time in half. Ask AI to apply that only to drafting: twenty minutes of drafting plus twenty minutes of review equals forty minutes per brief before any new work is added. Do not halve the entire sixty-minute task when the assumption covers only one component.
Add the work needed to use the output
An assisted workflow may require checking or preparation that the original estimate omitted.
The exercise adds ten minutes per brief for checking AI-generated claims. The modeled assisted total becomes fifty minutes per brief, or five hundred minutes for ten briefs. The modeled difference is one hundred minutes, not three hundred. State that the result follows from the supplied assumptions.
Keep input preparation, correction and supervision visible if they become part of the scenario. Do not count review time twice under different labels, and do not omit it because it is performed by another role. The worksheet should show where each minute belongs and who would need to verify the assumption.
Test when the modeled saving disappears
A sensitivity case helps identify which assumption matters before collecting data.
If checking requires twenty minutes rather than ten, the assisted total returns to sixty minutes per brief. Under that scenario, the modeled time difference is zero. This does not establish that AI is ineffective; it shows that the conclusion depends on a checking-time assumption that needs evidence.
Ask AI to create the two cases without assigning invented probabilities. Identify which task observations would help estimate drafting, checking and correction time. Keep task quality in the evidence plan so a shorter duration is not treated as improvement when the output fails its intended use.
Distinguish time released from usable capacity
A calculated saving does not automatically become a removable role or a larger output commitment.
One hundred minutes spread across ten briefs may occur at different times and for different workers. The fictional model does not establish that the time can be combined, reassigned or used to meet another deadline. It also says nothing about demand variation, coverage requirements or specialist responsibilities.
Prepare questions about scheduling and alternative work rather than translating the minutes into a headcount reduction. Keep decisions about people outside the calculation and with authorized human processes. The exercise is a way to challenge an operational assumption, not a tool for ranking employees or automating staffing decisions.
Deliver an assumption-led evidence request
The final worksheet should make the next measurement question explicit.
Show the six-hundred-minute baseline scenario, five-hundred-minute assisted scenario and zero-saving sensitivity case. List which inputs are hypothetical and what observations would be needed before using them in a real plan. Include task quality, correction burden and the timing of any released capacity.
Review whether the worker applied the change to the correct component and resisted the leap from time to staffing. The operations manager guide schedules a known workload; this guide examines whether the workload assumptions remain credible when an AI-assisted step is introduced.
Sources and scope
NIST: AI risk management. Skills England: workplace AI foundations.
These references provide background, not validation or endorsement of this exercise. The case details, calculations and suggested review questions are original instructional material. Use them to discuss observable work, not to infer customer outcomes, professional credentials or performance in every setting. Before adapting the exercise, confirm the relevant facts, approved tools, data permissions and decision owners. If you change the case, revisit the expected answers and checks as well. These examples describe practice tasks, not a promise that a particular product includes the fictional features.
Sources: [1] [2]
Sources
- 1.AI RMF Core · NIST
- 2.AI foundation skills for work benchmark · Skills England