AI Implementation, Maintenance, and Decommissioning AAIR Practice Questions (Page 7)

Questions 31–32

1. 31application · medium

   A telecom company is updating its AI-based network anomaly detection model. The new model has been trained on recent data and shows improved accuracy in testing. The company wants to ensure that the update does not introduce new false positives that could overwhelm the network operations team. What should they do before full deployment?

   • ATrain the new model with a higher false positive threshold to reduce alerts.
   • BDeploy the new model immediately and monitor the alert volume for a week.
   • CReview the new model's precision and recall on the test set only.
   • DRun the new model in a shadow mode alongside the current model, comparing alerts for a defined period.
   Reveal Solution Discussion Select an answer first
2. 32foundation · easy

   What is the purpose of validating an AI model against a separate test dataset during implementation?

   • ATo reduce the model's inference latency
   • BTo assess the model's performance on data it has not seen during training
   • CTo ensure the model is compatible with all operating systems
   • DTo increase the size of the training dataset
   Reveal Solution Discussion Select an answer first