AI Risk Metrics, Monitoring, and Reporting AAIR Practice Questions (Page 2)

Questions 6–10

1. 6foundation · easy

   Which of the following is an example of a technical AI risk metric?

   • AModel accuracy on a validation dataset
   • BPercentage of budget spent on AI governance
   • CEmployee satisfaction with AI tools
   • DNumber of data privacy complaints received
   Reveal Solution Discussion Select an answer first
2. 7application · medium

   After a quarterly review, the AI risk team identified a rising trend in false positives from a recommendation system. What is the most appropriate action to support informed decision-making?

   • AWait until the next quarterly review to see if the trend continues
   • BImmediately shut down the recommendation system
   • COnly inform the technical team and not management
   • DDocument the trend and communicate the findings to relevant stakeholders with a recommendation for further investigation
   Reveal Solution Discussion Select an answer first
4. 8foundation · easy

   Which of the following is a key component of a continuous monitoring process for AI systems?

   • AConducting annual audits of AI systems
   • BPerforming a one-time risk assessment at deployment
   • CReviewing AI risk metrics only after an incident occurs
   • DDefining data collection methods and frequency
   Reveal Solution Discussion Select an answer first
5. 9application · medium

   A financial services firm wants to automate the monitoring of its AI trading algorithms. The monitoring system must capture real-time data on model inputs, outputs, and performance metrics, and alert the risk team when thresholds are breached. Which tool configuration best meets this requirement?

   • AA manual process where the risk team periodically queries the model logs
   • BA real-time streaming platform that ingests model events and triggers alerts based on predefined thresholds
   • CA static dashboard that displays historical metrics without alerting
   • DA batch job that runs nightly to compute metrics and send a daily email report
   Reveal Solution Discussion Select an answer first
6. 10application · medium

   An AI risk manager needs to report on a model's performance to the IT security team, who are concerned about adversarial attacks. What type of information should be emphasized in the report?

   • AThe model's processing speed and latency
   • BThe model's accuracy on the latest test set
   • CThe number of attempted adversarial attacks detected and the model's robustness to such attacks
   • DThe cost savings achieved by the AI system
   Reveal Solution Discussion Select an answer first