AI Risk Metrics, Monitoring, and Reporting AAIR Practice Questions (Page 6)

Questions 26–30

1. 26foundation · easy

   When mapping AI risk metrics to risk tolerance levels, what does a 'tolerance level' typically define?

   • AThe number of AI models that can be deployed
   • BThe minimum performance required for AI models
   • CThe budget allocated for AI risk management
   • DThe maximum acceptable deviation from a risk target
   Reveal Solution Discussion Select an answer first
2. 27application · medium

   An organization's AI risk team is developing a reporting framework for AI risks. To ensure consistency with the organization's overall risk management, what should they do?

   • AReport AI risks only to the IT department, not to the enterprise risk committee
   • BAlign the AI risk reporting with the enterprise risk management framework and use similar risk categories and reporting lines
   • CUse a different risk appetite for AI risks than for other enterprise risks
   • DCreate a separate reporting framework exclusively for AI risks
   Reveal Solution Discussion Select an answer first
4. 28expert · hard

   A financial institution uses AI for real-time fraud detection. The risk team needs to monitor the model's performance continuously, but the data volume is extremely high, and real-time processing is costly. What is the most cost-effective approach that still provides timely risk monitoring?

   • AOnly monitor the model when a fraud incident is reported
   • BProcess all data in real time using expensive streaming infrastructure
   • CUse a sampling approach to monitor a representative subset of transactions in real time
   • DBatch process the data daily and review the metrics the next day
   Reveal Solution Discussion Select an answer first
5. 29application · medium

   A bank's AI-based loan approval system has a documented risk appetite that tolerates a model error rate of up to 2% for approved loans. The risk team wants a metric that directly indicates whether the system is operating within this tolerance. Which metric should they define as the primary KRI?

   • AAverage time to process a loan application
   • BPercentage of approved loans that default within 12 months
   • CNumber of customer complaints about the loan process
   • DNumber of model retraining events per quarter
   Reveal Solution Discussion Select an answer first
6. 30foundation · easy

   Why is it important to document AI risk monitoring results?

   • ATo ensure that all stakeholders receive the same report
   • BTo provide a historical record that supports trend analysis and decision-making
   • CTo replace the need for real-time monitoring
   • DTo satisfy the requirement that all AI risks be eliminated
   Reveal Solution Discussion Select an answer first