Data Management Specific to AI AAIA Practice Questions (Page 5)

Questions 21–25

1. 21application · medium

   A human resources department is using an AI model to screen job applicants. The model was trained on historical hiring data from the past 10 years. The data science team has discovered that the model is less likely to recommend female candidates for technical roles. The team wants to address this bias. Which initial step is most appropriate to investigate the root cause?

   • AAnalyze the training data to determine if there is a historical imbalance in the representation of female candidates for technical roles.
   • BReplace the model with a simpler rule-based system that does not use historical data.
   • CImmediately adjust the model's decision threshold to increase the number of female candidates recommended.
   • DRemove the gender feature from the model's input data and retrain the model.
   Reveal Solution Discussion Select an answer first
2. 22application · medium

   A media company is building an AI model to analyze video content. The training dataset is approximately 50 TB and is expected to grow by 10 TB per month. The data science team needs a storage solution that supports high-throughput reads for training and can scale cost-effectively. The data is not accessed frequently after the initial training run. Which storage architecture is most appropriate for this workload?

   • AUse a relational database (e.g., PostgreSQL) to store the video files as binary large objects (BLOBs).
   • BUse a cloud object storage service (e.g., Amazon S3 or Azure Blob Storage) for the raw data, and load the data into a high-performance compute cluster during training.
   • CStore the data in a Hadoop Distributed File System (HDFS) cluster that is always running.
   • DStore all data on a network-attached storage (NAS) device in the company's on-premises data center.
   Reveal Solution Discussion Select an answer first
4. 23foundation · easy

   Which security control is used to protect sensitive data in a non-production environment by replacing it with fictional but realistic values?

   • ANetwork segmentation
   • BMulti-factor authentication
   • CEncryption at rest
   • DData masking
   Reveal Solution Discussion Select an answer first
5. 24foundation · easy

   Which of the following best describes data provenance?

   • AThe method of backing up data to prevent loss
   • BThe documented origin and history of a dataset, including its source and any changes made
   • CThe technique of splitting data into training and testing sets
   • DThe process of encrypting data before storage
   Reveal Solution Discussion Select an answer first
6. 25expert · hard

   A large enterprise is adopting a federated data governance model for its AI initiatives. Each business unit has its own data governance team, but the enterprise wants to ensure consistent AI data management practices across all units. The central AI office has issued a set of principles, but the business units are interpreting them differently. Which governance mechanism is most effective in reconciling these differences while respecting the federated model?

   • AAllow each business unit to operate independently and only report their AI data practices annually to the central office.
   • BMandate that all business units adopt the exact same data management policies, overriding local governance teams.
   • CCreate a central data catalog that all business units must use, but allow them to define their own quality rules.
   • DEstablish a central AI data governance council with representatives from each business unit to review and approve unit-level policies for consistency.
   Reveal Solution Discussion Select an answer first