Executive AI Education: Use a Decision Rehearsal, Not a Feature Tour

Choose a decision the executive can actually make

Select one realistic choice: approve a pilot, expand a workflow, respond to an incident, revise a policy, or request stronger evidence. Define the available information, time limit, affected stakeholders, and authority boundary. Avoid a fictional scenario so dramatic that participants can succeed by giving generic caution rather than weighing evidence.

NIST’s AI RMF is intended for voluntary, use-case-aware management of AI risks and organizes work through govern, map, measure, and manage (NIST AI RMF). Those functions offer a useful sequence for executive questions without turning the session into a certification exercise.

Sources: [1]

Build an executive decision packet

Include a one-page business case, workflow map, evidence summary, risk register, stakeholder notes, and two deliberate contradictions. Each source is dated and its limitations are visible. The facilitator holds back one update until participants have stated their initial decision criteria. This reveals whether new evidence changes the recommendation or is ignored.

GAO’s AI accountability framework emphasizes governance, data, performance, and monitoring and provides questions for managers and assessors (GAO AI accountability framework). The packet uses those lenses to prevent a business case from focusing only on expected benefit while omitting data fitness, operating performance, and continuing oversight.

Sources: [2]

Worked rehearsal: expand or hold

Imagine a synthetic internal drafting pilot with strong adoption, favorable user comments, and a reported reduction in turnaround time. The packet also shows inconsistent quality sampling, a new data connection, and no named owner for external claims. Halfway through, participants learn that the logging definition changed during the comparison period.

The executive team must choose expand, restrict, repair, or hold. A strong decision distinguishes adoption from quality, asks for comparable evidence, assigns the data and claims owners, and limits the next phase to approved information. The record captures the criteria, dissent, conditions, owner, review date, and evidence that would reverse the choice.

Debrief observable governance behavior

Review which sources were used, which contradictions were surfaced, how uncertainty was communicated, and whether authority was assigned. Do not score rhetorical confidence. Ask participants to rewrite one decision condition and explain the follow-up evidence. Compare the reasoning with the organization’s actual governance, not an abstract ideal.

Carry one action into operating practice

Close with a named change to a real decision forum, evidence packet, or escalation route. Check later whether it occurred and whether it improved clarity. Attendance and satisfaction remain learning signals, not proof of executive capability or programme impact.

Make dissent useful

Assign one participant to present the strongest case against the preferred decision using the same evidence packet. Another participant identifies what additional evidence would change their own view. The facilitator records whether disagreement comes from facts, risk tolerance, authority, or timing. This creates a better debrief than asking whether everyone enjoyed the exercise.

Repeat the rehearsal with a different decision type rather than reusing the same answer. A leader who handles a procurement expansion well may still need practice with incident response or workforce evidence. Report observed behavior by scenario and avoid combining unlike exercises into a precise enterprise ranking.

Keep facilitator notes separate from the participant-facing decision record. Record the scenario version, evidence revealed, timing, and any coaching provided. Without that context, comparisons across sessions can overstate differences that came from the exercise rather than executive judgment.

This educational method does not certify executives, establish mature benchmarks, or prove business outcomes. Decisions remain with authorized leaders using current organizational evidence and policy.