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 4)

Questions 16–20

1. 16foundation · easy

   Which privacy principle in AI-assisted data analysis requires that only the minimum amount of personal data necessary for the analysis be collected and processed?

   • AData retention
   • BPurpose limitation
   • CData minimization
   • DData sovereignty
   Reveal Solution Discussion Select an answer first
2. 17expert · hard

   An AI model used by a court to set bail amounts is found to have a higher false-positive rate for a particular demographic group. The court is considering whether to continue using the model. What is the most appropriate action to balance ethical and legal considerations?

   • AConduct a thorough fairness audit and, if bias is found, implement mitigation techniques or discontinue the model if it cannot be made fair.
   • BPublish the model's false-positive rates by demographic group to the public.
   • CContinue using the model because it is more accurate than human judges overall.
   • DDiscontinue the model and return to human-only bail decisions.
   Reveal Solution Discussion Select an answer first
4. 18application · medium

   A bank develops an AI model to approve loan applications. The model is trained on historical data where a protected attribute (e.g., race) was a strong predictor of default. The bank wants to deploy the model while adhering to ethical principles. What is the most appropriate action?

   • AEvaluate the model for disparate impact across protected groups and apply fairness mitigation techniques if bias is detected.
   • BDeploy the model as-is because the historical data reflects real-world default rates.
   • CPublish the model's decision logic to all applicants to ensure transparency.
   • DRemove the protected attribute from the training data and deploy the model.
   Reveal Solution Discussion Select an answer first
5. 19application · hard

   An AI model used by a university to screen student applications is found to consistently rank applicants from certain high schools lower. The university's data science team is unsure if this is due to bias or actual academic performance differences. What is the most appropriate next step?

   • APublish the model's ranking criteria to all applicants.
   • BConduct a fairness audit to compare model outcomes against actual student performance data.
   • CRemove high school from the model's features and redeploy.
   • DImmediately disable the model and revert to manual review.
   Reveal Solution Discussion Select an answer first
6. 20foundation · easy

   Which control is used to protect data at rest and in transit by converting it into a format that is unreadable without a decryption key?

   • AData masking
   • BEncryption
   • CAnonymization
   • DData minimization
   Reveal Solution Discussion Select an answer first