Data Analytics (Aggregation, AI, Data Warehouse) CDPSE Practice Questions (Page 4)

Questions 16–20

1. 16foundation · easy

   What is the purpose of AI governance?

   • ATo maximize the speed of AI model training
   • BTo eliminate the need for data protection measures
   • CTo replace human decision-making with automated systems
   • DTo establish frameworks and practices that ensure AI systems are used responsibly and protect data
   Reveal Solution Discussion Select an answer first
2. 17application · medium

   A data warehouse stores customer data for a marketing analytics team. The team frequently runs ad-hoc queries. The privacy office is concerned about unauthorized access to sensitive fields. Which control would provide the most granular protection for specific columns?

   • APlace the warehouse on a separate network segment.
   • BImplement column-level security to restrict access to sensitive columns based on user roles.
   • CCreate a backup of the warehouse and store it offline.
   • DEncrypt the entire database with transparent data encryption.
   Reveal Solution Discussion Select an answer first
4. 18application · medium

   A data warehouse contains personally identifiable information (PII) for a customer analytics platform. The security team wants to ensure that data is protected both at rest and during query execution. Which combination of controls should be implemented?

   • AStore the warehouse on a separate physical server.
   • BEncrypt the data at rest and use encrypted connections for all client queries.
   • CMask all PII columns in the warehouse.
   • DUse role-based access control only.
   Reveal Solution Discussion Select an answer first
5. 19application · medium

   A city government is deploying an AI system to allocate social services resources. The system uses historical case data that may contain biases. The privacy office must ensure the system respects data protection principles and is accountable. Which governance measure should be implemented first?

   • APublish the AI system's source code on a public repository.
   • BPurchase a commercial AI platform with built-in privacy features.
   • CAnonymize all historical case data before training.
   • DEstablish an AI ethics review board to approve the system's design and deployment.
   Reveal Solution Discussion Select an answer first
6. 20expert · medium

   A hospital is deploying an AI system to predict patient readmission risk. The system uses electronic health records (EHR) containing sensitive health data. The hospital must comply with HIPAA and also ensure the model is transparent to clinicians. The data science team proposes using a complex deep learning model. The privacy officer is concerned about explainability. Which approach best balances privacy, transparency, and clinical utility?

   • AOutsource the model development to a third-party vendor to avoid liability.
   • BUse a simpler, interpretable model (e.g., logistic regression) with feature importance analysis.
   • CUse a deep learning model and provide post-hoc explanations using LIME.
   • DDeploy the deep learning model but keep the training data secret.
   Reveal Solution Discussion Select an answer first