4.3 Explain the Ethical and Privacy Considerations in AI-assisted Data Analysis, Including Controls to Prevent Data Exposure AI-TECHNICAL-PRACTITIONER Practice Questions (Page 3)

Questions 11–15

1. 11expert · hard

   A hospital uses an AI model to predict patient deterioration. The model is hosted on a cloud platform. The hospital's compliance team is concerned about data exposure during model inference. What is the most appropriate control to protect patient data during inference?

   • AUse a confidential computing environment where the model and data are processed in a hardware-based trusted execution environment (TEE).
   • BAnonymize the patient data before sending it to the model.
   • CUse a private endpoint to connect to the cloud AI service, ensuring traffic does not traverse the public internet.
   • DEncrypt the data in transit using TLS.
   Reveal Solution Discussion Select an answer first
2. 12foundation · easy

   In AI-assisted data analysis, which ethical principle ensures that there is a designated party responsible for the outcomes and impacts of the AI system's decisions?

   • AData minimization
   • BAccountability
   • CFairness
   • DTransparency
   Reveal Solution Discussion Select an answer first
4. 13foundation · easy

   In AI data analysis, which privacy consideration ensures that personal data collected for one purpose is not reused for unrelated purposes without additional consent or legal basis?

   • APurpose limitation
   • BAccess control
   • CData minimization
   • DAnonymization
   Reveal Solution Discussion Select an answer first
5. 14foundation · easy

   Which control is specifically designed to prevent data exposure by rendering personal data irreversibly unidentifiable, so that individuals cannot be re-identified?

   • AAudit logging
   • BAnonymization
   • CAccess control
   • DEncryption
   Reveal Solution Discussion Select an answer first
6. 15application · medium

   A healthcare analytics team uses an AI tool to analyze patient records for readmission risk. The team only needs diagnosis codes and length of stay, but the dataset includes full names, addresses, and Social Security numbers. The team lead wants to proceed with the analysis while minimizing privacy risk. What should the team do first?

   • AApply data minimization by extracting only the diagnosis codes and length-of-stay fields into a separate dataset for analysis.
   • BAnonymize the full dataset by removing names and addresses, keeping Social Security numbers for record linkage.
   • CObtain written consent from every patient whose data is in the dataset.
   • DEncrypt the entire dataset with AES-256 and proceed with the analysis on the encrypted data.
   Reveal Solution Discussion Select an answer first