Data Management Controls AAISM Practice Questions (Page 2)

Questions 6–10

1. 6application · medium

   A financial institution uses an AI model to approve loans. Regulators require the institution to explain any adverse decision. The model uses multiple data sources, including credit bureau data, internal transaction history, and demographic data. The institution needs to trace the data used for a specific decision back to its source. What should be implemented?

   • AA data classification scheme that labels each data source by sensitivity.
   • BData lineage documentation that records the origin, transformations, and movement of data through the AI pipeline.
   • CA data retention schedule that specifies how long each data source is kept.
   • DAn access control policy that restricts who can view the data.
   Reveal Solution Discussion Select an answer first
2. 7expert · hard

   A hospital uses AI to predict patient readmission risk. The model is trained on historical patient data, but the data quality team discovers that some fields have missing values and others have duplicate records. The team must decide which data quality improvement to prioritize to maximize model performance while minimizing effort. Which approach is most effective?

   • AUse a data lineage tool to trace the source of missing values and duplicates.
   • BFocus only on imputing missing values, as duplicates are rare and have minimal impact.
   • CRemove all duplicate records and impute missing values using the mean.
   • DFocus only on removing duplicate records, as missing values can be handled by the model.
   Reveal Solution Discussion Select an answer first
4. 8application · medium

   A healthcare AI project uses patient records to train a diagnostic model. The data includes direct identifiers, clinical notes, and lab results. The organization's governance framework requires that access to the training dataset be restricted to the smallest group of people who need it for their role. Which data classification level should be assigned to the training dataset?

   • AInternal
   • BPublic
   • CConfidential
   • DRestricted
   Reveal Solution Discussion Select an answer first
5. 9foundation · easy

   Which component is typically included in a data governance framework to ensure accountability for data quality?

   • AData backup tapes
   • BData encryption keys
   • CData visualization tools
   • DData stewards
   Reveal Solution Discussion Select an answer first
6. 10application · medium

   A bank uses AI to detect money laundering. The model uses transaction data that must be retained for 5 years per regulatory requirements. After 5 years, the data must be disposed of in a way that prevents reconstruction. Which disposal method is most appropriate?

   • ACryptographic erasure, where the encryption keys are destroyed.
   • BAnonymizing the data before disposal.
   • CDeleting the data from the database and removing backups.
   • DDegaussing the storage media.
   Reveal Solution Discussion Select an answer first