Objective 3.3 Identify and Mitigate Transparency and Explainability Risks CERTIFIED-ETHICAL-EMERGING-TECHNOLOGIST Practice Questions (Page 5)

Questions 21–23

1. 21expert · hard

   A financial services firm uses a machine learning model to detect fraud. The model is highly accurate but is a black box. Regulators require the firm to explain individual decisions to customers. The firm also wants to maintain a competitive advantage by keeping the model proprietary. Which mitigation strategy best satisfies both regulatory and business requirements?

   • AReplace the model with a simpler, interpretable model, even if it reduces fraud detection accuracy.
   • BUse LIME to generate local explanations for each decision, and keep the model's internal logic confidential.
   • CProvide customers with a generic statement that the model uses transaction data to detect fraud.
   • DOpen-source the model's code to demonstrate transparency.
   Reveal Solution Discussion Select an answer first
2. 22expert · hard

   A retail company uses a machine learning model to personalize marketing offers. The model was trained on data that includes customer demographics and purchase history. The company wants to provide customers with explanations for why they receive certain offers, but is concerned that explaining the model's logic could reveal trade secrets. Which approach best addresses this concern?

   • ARefuse to provide explanations, citing the need to protect trade secrets.
   • BPublish the model's source code to demonstrate transparency.
   • CUse model-agnostic explanations that describe the general factors influencing offers without revealing the model's proprietary logic.
   • DProvide customers with the exact data points used to generate their offers.
   Reveal Solution Discussion Select an answer first
3. 23application · medium

   A bank is required by regulators to provide customers with explanations for credit decisions. The bank's model is a complex ensemble that is difficult to interpret. The bank wants to meet regulatory requirements without sacrificing model performance. Which mitigation strategy is most appropriate?

   • AReplace the ensemble with a simple logistic regression model.
   • BRefuse to provide explanations, citing the model's complexity.
   • CUse model-agnostic explanations, such as LIME or SHAP, to generate post-hoc explanations for each decision.
   • DProvide customers with the model's source code.
   Reveal Solution Discussion Select an answer first