AI Incident Response and Business Continuity CRAGE Practice Questions (Page 7)

Questions 31–35

1. 31application · medium

   An e-commerce company's AI-based product recommendation system depends on a real-time clickstream data pipeline. The pipeline is hosted on a single cloud instance. The company is developing a business continuity plan (BCP) to ensure the recommendation service remains available during a pipeline failure. What should the BCP include?

   • AA manual process for customer service representatives to provide recommendations.
   • BA backup of the model weights stored on a separate disk.
   • CA redundant data pipeline in a different availability zone and a fallback rule-based recommendation engine.
   • DA plan to increase the instance size of the current pipeline when failure occurs.
   Reveal Solution Discussion Select an answer first
2. 32foundation · easy

   Which stakeholder should be notified first during an AI incident that affects user data privacy?

   • AThe model's training data providers
   • BThe organization's marketing team
   • CThe general public via social media
   • DThe organization's data protection officer (DPO)
   Reveal Solution Discussion Select an answer first
4. 33application · medium

   A financial services firm's fraud-detection model suddenly begins flagging a high volume of legitimate transactions as fraudulent. The on-call data scientist suspects data drift but cannot confirm it yet. The incident response team has been activated. What should the team do FIRST to manage this incident effectively?

   • AShut down the model serving endpoint to prevent further false positives.
   • BTriage the incident by assessing the scope of affected transactions and the potential financial impact, then escalate to the incident commander.
   • CImmediately roll back the model to the previous production version to stop the false positives.
   • DNotify all customers that their transactions may be delayed due to a system issue.
   Reveal Solution Discussion Select an answer first
5. 34foundation · easy

   What is the primary purpose of a communication protocol for AI incidents?

   • ATo assign blame to the responsible team
   • BTo ensure timely and accurate information is shared with the right stakeholders
   • CTo prevent any information from being shared externally
   • DTo delay notification until the incident is fully resolved
   Reveal Solution Discussion Select an answer first
6. 35application · medium

   After a ransomware attack encrypted the data used by a hospital's patient diagnosis AI, the IT team restored the data from backups. The AI model is now running, but the hospital wants to ensure it can recover quickly from future incidents. What should they do as part of recovery?

   • AImmediately resume normal operations without testing the restored model.
   • BTest the restored model and data to ensure they function correctly, and update the recovery procedures based on lessons learned.
   • CDelete the backups to free up storage space.
   • DSwitch to a manual diagnosis process permanently.
   Reveal Solution Discussion Select an answer first