AI Capabilities, Data Dependencies, and Failure Modes CAIPM Practice Questions (Page 3)

Questions 11–15

1. 11foundation · easy

   To mitigate bias in an AI hiring tool, which strategy is most directly aimed at addressing the root cause?

   • AImplementing robust data governance
   • BDeploying the model to more users
   • CReducing the frequency of model retraining
   • DIncreasing the model's learning rate
   Reveal Solution Discussion Select an answer first
2. 12foundation · easy

   A manufacturing company wants to use AI to inspect products on an assembly line for visual defects. Which core AI capability is most essential for this task?

   • AReasoning
   • BNatural language processing
   • CLearning
   • DPerception
   Reveal Solution Discussion Select an answer first
4. 13expert · hard

   A ride-sharing company uses an AI model to predict demand for rides in different neighborhoods. The model was trained on data from the past year. After a major event changes the city's traffic patterns, the model's predictions become less accurate. The company's data science team is small and cannot retrain the model frequently. What is the most sustainable approach?

   • ASet up automated monitoring and a scheduled retraining pipeline that uses recent data.
   • BFreeze the model and rely on human dispatchers to correct predictions.
   • CUse a simpler model that is less sensitive to changes.
   • DManually adjust the model's parameters whenever predictions are off.
   Reveal Solution Discussion Select an answer first
5. 14application · medium

   A hospital is developing an AI model to predict patient readmission risk from electronic health records (EHR). The training data is drawn from a single hospital's records over the past three years. The model performs well on historical data but is expected to be deployed across multiple hospitals in different regions. Which data dependency is most likely to cause a failure mode, and what is the best mitigation?

   • AData relevance; the model may not generalize to different patient populations, so validate on multi-site data and retrain with regional data.
   • BData availability; the model needs real-time data feeds, so set up streaming pipelines to each hospital.
   • CData quantity; the model needs more rows from the same hospital, so collect additional historical records from that site.
   • DData quality; the EHR contains missing values, so impute them with the mean before deployment.
   Reveal Solution Discussion Select an answer first
6. 15application · medium

   A manufacturing company uses an AI system to predict equipment failures from sensor data. The system's accuracy is high, but the company wants to ensure it remains reliable as machinery ages and operating conditions change. Which strategy is most important to maintain performance over time?

   • AFreeze the model to ensure consistent predictions.
   • BIncrease the model's complexity to capture more patterns.
   • CReplace the model with a rule-based system based on expert knowledge.
   • DContinuously monitor model performance and retrain with new sensor data periodically.
   Reveal Solution Discussion Select an answer first