Integrating AI into Enterprise Systems CAIPM Practice Questions (Page 7)

Questions 31–35

1. 31application · medium

   A media company has integrated an AI model for content recommendation on its streaming platform. The model is updated weekly with new training data. The company wants to ensure that the model's recommendations remain relevant and that any performance degradation is detected quickly. What should the monitoring plan include?

   • AAutomatically retrain the model every day with the latest data, regardless of performance.
   • BMonitor only the system's uptime and API response times.
   • CReview the model's training data manually each week to ensure it is accurate.
   • DTrack user engagement metrics (e.g., click-through rate) and compare them against a baseline to detect model drift.
   Reveal Solution Discussion Select an answer first
2. 32application · medium

   A multinational corporation is integrating a customer-churn prediction model. Customer data resides in separate CRM and billing systems in different regions, with inconsistent formats and duplicate records. What should the data integration plan include first?

   • ABuild a data lake that stores raw data from all sources without transformation.
   • BUse the CRM data as the primary source and ignore the billing system data.
   • CCreate a single customer master record by standardizing formats and deduplicating records across sources.
   • DTrain the model separately on each regional dataset and combine the predictions.
   Reveal Solution Discussion Select an answer first
4. 33foundation · easy

   Which practice is most important for ensuring high-quality data for AI models during data integration?

   • AUsing only data from internal sources to avoid external data issues
   • BStoring all data in its raw format without any transformation
   • CMinimizing the number of data sources to reduce complexity
   • DData profiling and cleansing to identify and correct inconsistencies
   Reveal Solution Discussion Select an answer first
5. 34application · medium

   An insurance company has deployed an AI underwriting model on-premises. The model's performance has degraded over time because customer behavior has changed. The program manager needs to detect this drift and retrain the model. What should be established?

   • AA monthly manual review of the model's outputs by underwriters.
   • BAutomated monitoring of prediction accuracy against actual outcomes, with alerts when accuracy drops below a threshold.
   • CA dashboard that shows the number of predictions made each day.
   • DA scheduled retraining job that runs every quarter regardless of performance.
   Reveal Solution Discussion Select an answer first
6. 35application · medium

   A university is integrating an AI-based student advising system that recommends courses based on academic history and career goals. Faculty advisors are concerned that the system will replace their roles. The university wants to ensure advisors use the system as a tool. What should the change management plan include?

   • ALet advisors opt out of using the system if they prefer not to.
   • BProvide a one-time email explaining the benefits of the system to advisors.
   • CDeploy the system immediately and require advisors to use it for all advising sessions.
   • DInvolve faculty advisors in the design and pilot phases, and clearly communicate that the system augments their role, not replaces it.
   Reveal Solution Discussion Select an answer first