Objective 3.3 Identify and Mitigate Transparency and Explainability Risks CERTIFIED-ETHICAL-EMERGING-TECHNOLOGIST Practice Questions (Page 2)

Questions 6–10

1. 6application · medium

   A data science team is developing a credit scoring model and needs to document its behavior for internal auditors. They want to provide both global and local explanations of the model's predictions. Which combination of tools would best achieve this?

   • AUse LIME for both global and local explanations.
   • BUse LIME for global explanations and SHAP for local explanations.
   • CUse a model card for local explanations and SHAP for global documentation.
   • DUse SHAP for local explanations and a model card for global documentation.
   Reveal Solution Discussion Select an answer first
2. 7expert · hard

   A healthcare organization is deploying an AI system to assist with diagnosis. The system was trained on data from a single hospital and may not generalize well to other populations. The organization wants to ensure that the system is transparent to both clinicians and patients, but also wants to avoid overburdening clinicians with excessive information. Which approach best addresses these competing needs?

   • AProvide clinicians with detailed technical explanations using SHAP, and give patients a simplified summary of the system's purpose and limitations.
   • BProvide no explanations to patients to avoid confusion, and only brief summaries to clinicians.
   • CPublish the model's source code and training data for public review.
   • DProvide the same detailed technical explanation to both clinicians and patients.
   Reveal Solution Discussion Select an answer first
4. 8foundation · easy

   Which tool is specifically designed to provide local explanations for individual predictions of a machine learning model?

   • ATensorFlow
   • BDocker
   • CModel cards
   • DLIME
   Reveal Solution Discussion Select an answer first
5. 9expert · hard · select all that apply

   A tech company is developing an AI system for hiring. The system uses a complex model that may have biases. The company wants to identify and mitigate transparency and explainability risks before deployment. Which of the following actions would be appropriate? Select all that apply.

   • ALimit the use of the system to a small group of HR professionals.
   • BConduct a model documentation audit to review the system's design and data sources.
   • CImplement a governance policy that requires explanations for hiring decisions.
   • DUse SHAP to analyze feature contributions and detect potential biases.
   • EKeep the model's logic confidential to avoid legal challenges.
   Check answers Discussion Select an answer first
6. 10foundation · easy

   Which mitigation strategy is most effective for reducing transparency risks when an AI model is too complex for stakeholders to understand directly?

   • AUsing model-agnostic explanations
   • BDeleting the model's training data
   • CRestricting access to the model's outputs
   • DIncreasing the model's complexity
   Reveal Solution Discussion Select an answer first