Objective 3.3 Identify and Mitigate Transparency and Explainability Risks CERTIFIED-ETHICAL-EMERGING-TECHNOLOGIST Practice Questions (Page 3)

Questions 11–15

1. 11application · medium

   A team is building a machine learning model to detect fraudulent transactions. They need to explain individual decisions to customers who may be affected. Which tool would be most appropriate for generating explanations for each flagged transaction?

   • AModel cards, because they document the model's overall performance.
   • BLIME, because it provides local explanations for individual predictions.
   • CData provenance logs, because they track the origin of training data.
   • DSHAP, because it provides global feature importance across all predictions.
   Reveal Solution Discussion Select an answer first
2. 12foundation · easy

   A company deploys an AI system to approve loan applications. The system's decisions are based on data collected from various sources, but the company does not clearly communicate to applicants how their data is used. Which transparency or explainability risk is most directly present?

   • AAlgorithmic opacity
   • BModel performance degradation
   • CStakeholder information asymmetry
   • DData provenance issues
   Reveal Solution Discussion Select an answer first
4. 13application · medium

   A hospital uses an AI system to prioritize patient triage. The system was trained on historical data that may contain biases. The hospital wants to ensure that both clinicians and patients understand the system's limitations. Which mitigation strategy is most effective?

   • AProvide clinicians with raw training data to review.
   • BRestrict access to the AI system to senior clinicians only.
   • CDisable the AI system and rely on manual triage.
   • DPublish a model card that documents the system's intended use, performance, and known limitations.
   Reveal Solution Discussion Select an answer first
5. 14application · medium

   A tech company is deploying an AI system for resume screening. The HR team is concerned about potential bias and wants to document the system's characteristics for external audit. Which tool would best support this documentation effort?

   • ASHAP, because it identifies feature importance.
   • BLIME, because it explains individual predictions.
   • CModel cards, because they provide a structured summary of the model's purpose, performance, and limitations.
   • DData provenance logs, because they track data lineage.
   Reveal Solution Discussion Select an answer first
6. 15expert · hard

   A company uses a machine learning model to predict customer churn. The model is a gradient boosting machine, and the company wants to provide explanations to customer service representatives so they can better understand why a customer is at risk. The representatives are not data scientists and need intuitive explanations. Which tool would be most appropriate?

   • ALIME, because it generates local explanations that are easy to understand.
   • BSHAP, because it provides global feature importance.
   • CModel cards, because they document the model's overall performance.
   • DData provenance logs, because they track the origin of training data.
   Reveal Solution Discussion Select an answer first