Flexibility and Adaptability CT-AI Practice Questions (Page 2)

Questions 6–10

1. 6application · medium

   A fraud detection system for a bank was trained on transaction data from 2020. Since then, fraud patterns have evolved significantly, and the system's false positive rate has increased. The bank wants the system to maintain its detection accuracy. Which approach best addresses the need for adaptability?

   • AFreeze the model and rely on manual review of all flagged transactions.
   • BReplace the model with a simpler rule-based system.
   • CRegularly retrain the model with new transaction data and updated fraud labels.
   • DIncrease the model's decision threshold to reduce false positives.
   Reveal Solution Discussion Select an answer first
2. 7foundation · easy

   An AI-based traffic management system adjusts traffic light timings in real time based on current traffic flow. This is an example of:

   • AAdaptability, because the system learns from historical traffic data.
   • BRobustness, because the system continues to operate despite sensor failures.
   • CFlexibility, because the system changes its outputs in response to changing traffic conditions.
   • DExplainability, because the system provides reasons for its timing decisions.
   Reveal Solution Discussion Select an answer first
4. 8expert · hard

   A credit scoring model is used by a bank to approve loans. The model is updated monthly with new applicant data. However, the bank's regulators require that the model's decisions be explainable. After an update, the model's feature importance changes, making some previously approved applicants now rejected. The bank is concerned about the trade-off between adaptability and explainability. Which approach best balances these concerns?

   • AUse a simpler model that is inherently explainable, even if it is less adaptable.
   • BRetrain the model only when absolutely necessary to minimize explanation changes.
   • CFreeze the model to ensure explanations remain stable.
   • DMonitor the model's explanations after each update and document changes for regulatory review.
   Reveal Solution Discussion Select an answer first
5. 9application · medium

   A predictive maintenance system for industrial equipment uses sensor data to predict failures. The equipment models change over time as new versions are released. The system must maintain its prediction accuracy. Which approach best supports the system's adaptability?

   • ATest the model with historical sensor data only.
   • BContinuously retrain the model with new sensor data from the latest equipment versions.
   • CIncrease the model's sensitivity to detect more failures.
   • DKeep the model unchanged to ensure consistent predictions.
   Reveal Solution Discussion Select an answer first
6. 10application · medium

   A recommendation system for an e-commerce site uses a model that is updated weekly with new purchase data. The team wants to verify that the model's recommendations improve after each update. Which testing approach is most appropriate?

   • ATest the model with the original training data to ensure it still works.
   • BRun a one-time evaluation with a fixed test set.
   • CCompare the model's performance on recent data before and after each update.
   • DCheck the model's explainability for its recommendations.
   Reveal Solution Discussion Select an answer first