How to Use This Module
For each scenario: read it, form your own response, then read the analysis. The analysis is not "the answer" — it's a structured way of thinking about the key tensions and trade-offs. Notice where your response aligned with the analysis and where it diverged, and ask yourself why.
The five scenarios are drawn from patterns that appear repeatedly in AI leadership practice. The specific contexts are composite; the decisions are real.
Scenario 1: The Speed vs. Governance Dilemma
Your competitor has deployed an AI-assisted customer service system that is clearly outperforming yours. Your team can deploy a comparable system in 8 weeks, but your governance process requires a full AI risk review that takes 12 weeks. The business case for moving fast is clear. The governance process exists because of a real incident 18 months ago. What do you do?
Key tensions: Competitive necessity vs. governance integrity. Precedent risk if you bypass process vs. competitive cost if you don't. The signal sent to the organisation about whether governance is real or performative.
Analysis: The worst response is quietly bypassing the governance process without acknowledging you're doing it. This destroys the credibility of future governance. Better options: convene an emergency governance review with clear scope limitations; delay one component of deployment while fast-tracking the review; be transparent with the board about the trade-off and get explicit authorisation for a time-limited exception. The governance process either matters or it doesn't — the leadership choice is which.
Scenario 2: The Workforce Displacement Announcement
An AI deployment you've led has reduced the headcount requirement in one function by 30%. You knew this was a probable outcome when you approved the project. The affected employees don't know yet. How do you communicate this, and to whom, and when?
Key tensions: Honesty vs. operational continuity. Legal constraints vs. ethical obligations. The signal this sends about how future AI deployments will be handled.
Analysis: The sequencing matters as much as the message. Telling the board before you tell affected employees, or communicating via HR without personal leadership involvement, will be remembered. The minimum standard: affected employees hear this from their direct leadership, not from HR, not via announcement. The substance: honesty about what happened, what support is available, and what the organisation's posture toward affected individuals is. Leaders who handle this with genuine care for individuals build the trust that enables future AI deployment; leaders who handle it procedurally erode it.
Scenario 3: The Rogue Excellence Problem
One team in your organisation has developed highly sophisticated AI workflows entirely outside your official AI programme. They're getting significantly better results than the official programme. The problem: they've used an unapproved external AI tool with data handling implications that haven't been assessed. What do you do?
Key tensions: Rewarding innovation vs. enforcing policy. Data risk vs. capability loss if you shut them down. The signal to other teams about what happens to people who move faster than the process.
Analysis: This is a common pattern — the best AI practitioners often run ahead of governance. The worst response is immediate shutdown without acknowledgment of the value they've created. A better approach: conduct an expedited data risk assessment; engage the team as experts, not as rule-breakers; find a path to legitimate status for the good parts of what they've built; and update your programme to incorporate what they've learned. The team should understand that the process exists for real reasons, but should also see that the organisation is capable of recognising and incorporating genuine innovation.
Scenario 4: The Board Sceptic
A board member with significant influence is vocal in their scepticism about AI investment — they believe it's hype, cite well-publicised AI failures, and have created enough uncertainty that other board members are asking questions about your AI strategy. A board presentation is coming up. How do you approach it?
Analysis: Sceptical board members often have legitimate concerns dressed in generalised doubt. The wrong approach is a polished AI showcase that ignores their concerns. Better: request a pre-meeting conversation to understand the specific concerns; structure the board presentation to address them directly rather than around them; present actual results with honest failure acknowledgment alongside success; and resist the temptation to oversell. A board member converted from sceptic to constructive questioner is more valuable than a board member who is performing enthusiasm they don't feel.
Scenario 5: The External AI Failure Attribution
An AI system your organisation deployed has produced outputs that caused measurable harm to a customer. The harm is real but not catastrophic. The AI was operating within its documented parameters. Who is responsible? How do you respond internally and externally?
Analysis: "It was within parameters" is not a sufficient public response — it sounds like deflection and probably is. The operational questions: what did we know, when did we know it, and what did we do with what we knew? The leadership question: does taking responsibility create more trust long-term than defending the deployment decision? In most cases: honest acknowledgment of the outcome, genuine investigation of what happened, and visible remediation create more durable trust than defensive posturing — with customers, regulators, and employees.