Objective 3.3 Identify and Mitigate Transparency and Explainability Risks CERTIFIED-ETHICAL-EMERGING-TECHNOLOGIST Practice Questions (Page 1)

Questions 1–5

1. 1foundation · easy

   An organization wants to identify transparency risks in its AI system by reviewing whether the system's design and development processes are adequately documented. Which method is most appropriate?

   • AAdversarial testing
   • BBias audit
   • CExplainability testing
   • DModel documentation audit
   Reveal Solution Discussion Select an answer first
2. 2expert · hard

   A multinational corporation uses an AI system to recommend personalized content to users. The system was trained on data from multiple countries, and the model is a deep neural network. The legal team is concerned about compliance with data protection regulations that require meaningful explanations for automated decisions. The company also wants to maintain the model's performance. Which approach best balances explainability and performance?

   • AReplace the deep neural network with a decision tree to ensure inherent interpretability.
   • BLimit the system's use to countries where explainability is not required.
   • CProvide users with a general statement that the system uses personal data to personalize content, without specific explanations.
   • DImplement SHAP to generate local explanations for each recommendation, and document the model's limitations in a model card.
   Reveal Solution Discussion Select an answer first
4. 3application · medium

   A marketing firm uses a machine learning model to segment customers for targeted campaigns. The firm's clients are asking for explanations of why certain customers are placed in specific segments. The firm wants to validate that the model's decisions are reasonable before communicating with clients. Which approach would best help them assess the model's explainability?

   • AReview the model's accuracy on a test set.
   • BConduct a stakeholder impact assessment to identify affected parties.
   • CUse SHAP to identify which features are most influential for each segment.
   • DCheck the model's training data for missing values.
   Reveal Solution Discussion Select an answer first
5. 4foundation · easy

   Which of the following is a primary source of transparency risk in an AI system?

   • AThe use of a well-known open-source machine learning library
   • BThe use of a simple linear regression model with a single feature
   • CA lack of documentation about the model's training data and feature engineering
   • DThe deployment of the model on a cloud platform with strong security controls
   Reveal Solution Discussion Select an answer first
6. 5foundation · easy

   Which technique is specifically used to detect whether an AI model's decisions are understandable to non-technical stakeholders?

   • AExplainability testing
   • BModel documentation audit
   • CStakeholder impact assessment
   • DData quality assessment
   Reveal Solution Discussion Select an answer first