AI Audit Outputs and Reports AAIA Practice Questions (Page 5)

Questions 21–25

1. 21foundation · easy

   Which adaptation is most important when presenting AI audit findings to a technical audience, such as data scientists and engineers?

   • ASummarize findings in a single page without supporting data
   • BAvoid all technical jargon to keep the report accessible
   • CFocus only on business impact and financial metrics
   • DProvide detailed evidence, model behavior specifics, and actionable technical recommendations
   Reveal Solution Discussion Select an answer first
2. 22application · medium

   During an AI audit of a customer service chatbot, the audit team found that the chatbot occasionally provides incorrect information about refund policies. The root cause is traced to outdated training data. Management wants a quick fix and proposes to manually correct the chatbot's responses for the most common refund questions. The audit team's recommendation should prioritize:

   • AImmediately retraining the model with the latest policy documents and redeploying it as soon as possible.
   • BDocumenting the finding in the report and recommending that the chatbot be taken offline permanently.
   • CAdding a disclaimer to the chatbot that responses may be inaccurate, to manage user expectations.
   • DImplementing a process to regularly update training data and adding a monitoring mechanism to detect outdated information.
   Reveal Solution Discussion Select an answer first
4. 23foundation · easy

   When formulating recommendations in an AI audit report, what should the auditor prioritize to ensure they address the root cause of findings?

   • AThe easiest fix that can be implemented quickly
   • BThe recommendation that requires the least resources regardless of effectiveness
   • CThe recommendation that aligns with organizational risk appetite and addresses the root cause
   • DThe recommendation that is most popular among stakeholders
   Reveal Solution Discussion Select an answer first
5. 24expert · hard

   An AI audit of a predictive maintenance model found that the model's accuracy varies significantly across different equipment types. The audit team is preparing the report for the maintenance manager, who is responsible for scheduling maintenance. Which approach best communicates the finding and supports decision-making?

   • ARecommend that the model be used only for equipment types where it is most accurate, and use manual inspections for others.
   • BPresent the accuracy variation by equipment type in a clear table, and recommend adjusting maintenance schedules based on the model's reliability for each type.
   • CProvide a detailed explanation of the model's algorithm and training data to justify the variation.
   • DState that the model is generally accurate and recommend no changes, as the variation is not significant.
   Reveal Solution Discussion Select an answer first
6. 25application · medium

   An AI audit of a supply chain forecasting model found that the model's accuracy is lower than expected during peak seasons. The audit team is preparing a report for the operations manager, who is responsible for inventory planning. Which presentation of the finding is most appropriate?

   • AExplain the accuracy drop in terms of potential stockouts or excess inventory, and recommend adjusting safety stock levels during peak seasons.
   • BRecommend replacing the model with a simpler one that is easier to understand, regardless of performance.
   • CProvide the model's mean absolute error and other statistical metrics for the operations manager to interpret.
   • DState that the model is generally accurate and recommend no changes, as the drop is only during peak seasons.
   Reveal Solution Discussion Select an answer first