Manage Accountability Documentation and Audit Trail PMI-CPMAI Practice Questions (Page 4)

Questions 16–20

1. 16expert · hard

   A data science team is training a model on a dataset that includes personally identifiable information (PII). The team has a chain of custody record that logs access to the data. However, a team member accidentally deletes a critical file from the dataset, and the team restores it from a backup. The chain of custody record does not reflect this incident. What is the most important action to take?

   • AUpdate the chain of custody record to document the deletion and restoration, including timestamps and the person responsible.
   • BDelete the backup and restore the original file from the cloud.
   • CNotify the data protection officer but do not update the records.
   • DIgnore the incident because the data was restored and no harm was done.
   Reveal Solution Discussion Select an answer first
2. 17foundation · easy

   Which record is a key part of maintaining chain of custody for a dataset?

   • AA log of who accessed the data, when, and for what purpose
   • BA list of all possible uses of the data in the future
   • CA description of the data's visual appearance
   • DA summary of the data's accuracy metrics
   Reveal Solution Discussion Select an answer first
4. 18foundation · easy

   What is the purpose of documenting stakeholder approvals at go/no-go decision points?

   • ATo provide a formal record that decisions were reviewed and authorized by the appropriate parties
   • BTo speed up the development process by skipping technical reviews
   • CTo create a marketing timeline for product launches
   • DTo replace the need for version control of the model
   Reveal Solution Discussion Select an answer first
5. 19application · medium

   A research team is using a dataset that was collected from a third-party vendor. The vendor requires that the team maintain a chain of custody record for the data. The team has stored the data in a shared folder and has a spreadsheet that logs who accessed the data and when. What is a limitation of this approach?

   • AThe spreadsheet can be easily edited or deleted, compromising the integrity of the chain of custody record.
   • BThe shared folder does not have enough storage space.
   • CThe spreadsheet does not include the data's content.
   • DThe spreadsheet is not accessible to all team members.
   Reveal Solution Discussion Select an answer first
6. 20expert · hard

   A company is developing an AI model that uses data from multiple sources, including some that are updated in real-time. The team needs to ensure that the model can be reproduced at any point in time. What is the most effective approach?

   • AUse a manual process to save a copy of the data whenever a model is trained.
   • BRecord the data source URLs and timestamps of access.
   • CStore only the final model weights and assume the data can be re-collected.
   • DImplement a data versioning system that captures snapshots of all data sources at regular intervals and links them to model training runs.
   Reveal Solution Discussion Select an answer first