AI Capabilities, Data Dependencies, and Failure Modes CAIPM Practice Questions (Page 4)

Questions 16–20

1. 16application · medium

   A bank's fraud detection model was trained on transaction data from 2022. In 2024, the model's false positive rate has increased significantly, and it is flagging many legitimate transactions. The bank's data scientists confirm the model's accuracy on the original test set is unchanged. What is the most appropriate first step to address this issue?

   • AIncrease the model's complexity to capture more patterns in the data.
   • BReplace the model with a rule-based system to avoid data dependencies.
   • CRetrain the model using recent transaction data and monitor its performance over time.
   • DReduce the false positive threshold to lower the number of alerts.
   Reveal Solution Discussion Select an answer first
2. 17application · medium

   A company's AI-based chatbot has been providing incorrect answers to customer questions after a recent update to the company's product line. The chatbot was trained on old product documentation. What is the most likely cause of the incorrect answers?

   • AThe training data is outdated and does not reflect the new product information.
   • BThe chatbot is overfitting to the training data.
   • CThe chatbot lacks natural language processing capabilities.
   • DThe chatbot is biased against certain customer questions.
   Reveal Solution Discussion Select an answer first
4. 18expert · hard

   A company wants to build an AI system that can read and summarize legal contracts. The system must also flag clauses that are likely to be problematic. Which combination of AI capabilities is most essential?

   • AReasoning to summarize and perception to identify clauses.
   • BPerception to scan the documents and learning to improve over time.
   • CNatural language processing for understanding text and reasoning to identify problematic clauses.
   • DLearning to adapt to new contract types and perception to read the text.
   Reveal Solution Discussion Select an answer first
5. 19application · medium

   A logistics company wants to use AI to predict delivery delays based on weather, traffic, and package volume. The company has historical data for the past two years, but the data is stored in different formats across regions and contains many missing values. What is the most critical data dependency to address before building the model?

   • AData quality; standardize formats and handle missing values to ensure consistent and reliable input.
   • BData relevance; ensure the data includes all possible delivery routes.
   • CData availability; set up real-time data feeds for weather and traffic.
   • DData quantity; collect more data from each region to ensure sufficient volume.
   Reveal Solution Discussion Select an answer first
6. 20application · medium

   A retail chain wants to use AI to forecast inventory demand for each store. The company has sales data for all stores, but some stores have only been open for a few months, while others have years of history. What is the main data dependency challenge for the new stores?

   • AData quantity; the new stores have insufficient historical data to train accurate models.
   • BData quality; the new stores' data is likely to be inaccurate.
   • CData relevance; the new stores' data is not relevant to the existing stores.
   • DData availability; the new stores do not have real-time data feeds.
   Reveal Solution Discussion Select an answer first