AML Communication & Stakeholder Relations 3 — Questions and Answers
Question 1: Which element is MOST critical when writing an executive summary for an ML project status report?
- Detailed hyperparameter tuning logs
- Key decisions needed, business impact achieved, and risks requiring attention (Correct answer)
- Full training and validation loss curves
- Complete API documentation for the model endpoint
Correct answer: Key decisions needed, business impact achieved, and risks requiring attention
Executive summaries must focus on decisions, impact, and risks because executives need to act on information, not understand technical implementation.
Question 2: A non-technical product manager claims the model is biased after seeing one incorrect prediction. How should you respond?
- Agree and immediately retrain the model
- Explain the difference between individual errors and systemic bias, and share aggregate fairness metrics (Correct answer)
- Dismiss the concern as statistically insignificant
- Transfer the manager to a different project team
Correct answer: Explain the difference between individual errors and systemic bias, and share aggregate fairness metrics
Distinguishing anecdotal errors from systemic bias using aggregate metrics educates stakeholders and prevents reactive decisions based on single observations.
Question 3: When managing stakeholder expectations for a new recommendation system, which approach prevents scope creep and misaligned expectations?
- Promise maximum performance to secure buy-in
- Define a shared project charter with measurable success criteria agreed upon by all stakeholders before development begins (Correct answer)
- Keep the project scope deliberately vague for flexibility
- Update stakeholders only when the model is fully deployed
Correct answer: Define a shared project charter with measurable success criteria agreed upon by all stakeholders before development begins
A shared project charter with pre-agreed success criteria creates accountability and prevents stakeholders from shifting expectations mid-project.
Question 4: Which communication medium is most appropriate for sharing real-time model monitoring alerts with operations teams?
- Quarterly PDF reports
- Automated dashboards with threshold-based alerting integrated into existing ops workflows (Correct answer)
- Annual performance reviews
- Informal verbal updates during stand-up meetings only
Correct answer: Automated dashboards with threshold-based alerting integrated into existing ops workflows
Automated dashboards with threshold alerts integrated into ops workflows ensure timely, actionable notifications without creating communication bottlenecks.
Question 5: A stakeholder group has conflicting interpretations of model output. What is the most effective resolution approach?
- Let each group use their own interpretation independently
- Convene a joint calibration session to standardize interpretation guidelines and document shared definitions (Correct answer)
- Choose the most senior stakeholder's interpretation automatically
- Remove the conflicting output from the model entirely
Correct answer: Convene a joint calibration session to standardize interpretation guidelines and document shared definitions
A calibration session with documented shared definitions eliminates ambiguity and ensures consistent model output interpretation across teams.
Question 6: When presenting A/B test results of two ML models to stakeholders, which statistical concept is most important to communicate clearly?
- The exact number of training epochs used
- Statistical significance and practical significance (effect size) and how they relate to business decisions (Correct answer)
- The model architecture details of each variant
- Server infrastructure costs of running each model
Correct answer: Statistical significance and practical significance (effect size) and how they relate to business decisions
Stakeholders must understand both statistical significance and practical effect size to avoid acting on results that are mathematically real but business-irrelevant.
Question 7: What is the primary purpose of a model card in stakeholder communication?
- To store model weights for deployment
- To provide structured documentation of model purpose, performance, limitations, and intended use for diverse audiences (Correct answer)
- To record hyperparameter search history
- To serve as a legal contract for model deployment
Correct answer: To provide structured documentation of model purpose, performance, limitations, and intended use for diverse audiences
Model cards provide standardized transparency documentation that helps diverse stakeholders — from engineers to policymakers — understand a model's appropriate use and limitations.
AML Communication & Stakeholder Relations 3
Which element is MOST critical when writing an executive summary for an ML project status report?