Objective 4.4 Address Business Risks, Ethical Concerns, and Related Concepts in Operationalizing the Model AIP-210 Practice Questions (Page 6)

Questions 26–29

1. 26foundation · easy

   What is a key element of actionable communication about ethical risks to stakeholders?

   • AIncluding specific mitigation steps and owners.
   • BAvoiding discussion of risks to prevent panic.
   • CProviding a list of all possible risks without prioritization.
   • DUsing statistical formulas to quantify risks.
   Reveal Solution Discussion Select an answer first
2. 27foundation · easy

   Why is ongoing monitoring of an ML model important after deployment?

   • ATo replace the model with a new one every week.
   • BTo ensure the model's performance remains stable over time.
   • CTo increase the model's complexity automatically.
   • DTo eliminate the need for human oversight.
   Reveal Solution Discussion Select an answer first
4. 28expert · hard

   A telecom company uses an ML model to predict customer churn. The model is used to target retention offers. The marketing team complains that the model's predictions are not actionable because it does not indicate why a customer is likely to churn. The data science team wants to improve the model's usefulness. What is the best approach?

   • ARetrain the model with more features to improve its predictive accuracy.
   • BReplace the model with a rule-based system that uses known churn indicators.
   • CAdd model explainability techniques, such as SHAP, to provide feature-level explanations for each prediction.
   • DProvide the marketing team with the model's confidence scores so they can prioritize high-confidence predictions.
   Reveal Solution Discussion Select an answer first
5. 29application · medium

   A hospital is deploying an ML model to predict patient readmission risk. The model uses demographic features including race and income. An internal audit reveals that the model has a higher false-negative rate for patients from lower-income neighborhoods. The hospital's ethics board asks the data science team to address this before launch. What is the most appropriate first action?

   • AAdjust the model's decision threshold for lower-income patients to reduce false negatives.
   • BCommunicate the finding to the ethics board and propose a fairness metric to monitor, then explore bias mitigation techniques.
   • CRemove race and income features from the model and retrain it.
   • DDelay the deployment indefinitely until the model is perfectly fair for all groups.
   Reveal Solution Discussion Select an answer first