AI Capabilities, Data Dependencies, and Failure Modes CAIPM Practice Questions (Page 1)

Questions 1–5

1. 1application · medium

   A retail company deploys an AI system that uses computer vision to count inventory on store shelves. The system performs well in stores with bright, uniform lighting but frequently miscounts items in stores with dim or mixed lighting. The company has abundant training images from the well-lit stores but very few from the dimly lit ones. What is the most likely cause of the performance gap?

   • AThe training data is not representative of the range of lighting conditions in the deployment environment.
   • BThe model is overfitting to the training data because it has too many parameters.
   • CThe computer vision model lacks natural language processing capabilities.
   • DThe model is suffering from data drift because customer behavior changed over time.
   Reveal Solution Discussion Select an answer first
2. 2application · easy

   A customer service department wants to deploy an AI system that can understand and respond to customer inquiries in multiple languages. Which combination of AI capabilities is most essential?

   • ANatural language processing and learning to improve responses over time.
   • BReasoning and perception to classify customer intent.
   • CLearning and perception to translate languages.
   • DPerception and reasoning to interpret customer emotions.
   Reveal Solution Discussion Select an answer first
4. 3application · medium

   A media company uses an AI system to recommend articles to users. The system was trained on user click data. After a major news event, the system starts recommending outdated articles because the training data does not include the event. What is the best way to handle this situation?

   • AManually curate the recommendations until the event is over.
   • BImplement a feedback loop that incorporates real-time user interactions and retrains the model frequently.
   • CIncrease the model's learning rate to make it adapt faster.
   • DFreeze the model to prevent it from making mistakes during the event.
   Reveal Solution Discussion Select an answer first
5. 4application · medium

   A telecom company uses an AI model to predict customer churn. The model was trained on data from the past year and is deployed to identify customers likely to churn in the next month. The company notices that the model's precision is high but recall is low; it misses many customers who actually churn. The business wants to reduce missed churners even if it means more false positives. What should the company do?

   • AUse a different algorithm that always predicts churn.
   • BLower the decision threshold for classifying a customer as a churn risk.
   • CRaise the decision threshold to be more conservative.
   • DRetrain the model with more features to improve precision.
   Reveal Solution Discussion Select an answer first
6. 5application · medium

   A manufacturing company uses an AI system to predict equipment failure. The system was trained on sensor data from machines. Over time, the sensors degrade and produce noisier readings. The company wants to maintain the system's accuracy without replacing all sensors immediately. What is the best approach?

   • ARetrain the model more frequently with the noisy data to adapt to the noise.
   • BImplement a data quality monitoring system to detect sensor degradation and recalibrate or replace sensors as needed.
   • CIgnore the noisy readings and use only historical data for predictions.
   • DUse a model that is robust to noisy data, such as a random forest.
   Reveal Solution Discussion Select an answer first