Describe the Impact of Predictive AI CERTIFIED-AI-EXPERT Practice Questions (Page 5)

Questions 21–25

1. 21expert · hard

   A bank uses a predictive model to approve loan applications. The model is accurate overall, but it denies loans to a higher proportion of applicants from a certain ethnic group. The bank must comply with fair lending regulations while maintaining profitability. What is the best approach?

   • AUse a generative model to create synthetic loan applications to balance the dataset
   • BRemove the ethnic group attribute from the model to eliminate bias
   • CIncrease the model's complexity to improve accuracy for all groups
   • DAudit the model for bias, retrain with balanced data, and monitor for disparate impact
   Reveal Solution Discussion Select an answer first
2. 22foundation · easy

   What is a key benefit of using predictive AI for risk management?

   • AIt guarantees compliance with all regulations.
   • BIt eliminates all financial losses.
   • CIt removes the need for human oversight in risk decisions.
   • DIt identifies potential risks early, allowing proactive mitigation.
   Reveal Solution Discussion Select an answer first
4. 23application · medium

   A financial institution has deployed a predictive model to detect fraudulent transactions. The model was trained on data from the past year. After six months, the fraud detection rate drops significantly. What is the most appropriate action to take?

   • AIncrease the model's complexity to capture more patterns
   • BFreeze the model to maintain consistent detection rates
   • CRetrain the model with recent transaction data to adapt to new fraud patterns
   • DSwitch to a generative AI model to create new fraud rules
   Reveal Solution Discussion Select an answer first
5. 24expert · hard

   An insurance company wants to build a predictive model to assess risk for auto insurance policies. They have historical claims data, but it is heavily skewed toward low-risk drivers. The company wants to avoid bias against high-risk drivers. What is the most important data-related action to take?

   • AEnsure the training data includes a representative sample of high-risk drivers
   • BUse a generative model to create synthetic low-risk drivers to balance the dataset
   • CUse a larger dataset even if it is skewed, because more data is always better
   • DAnonymize the data to protect driver privacy
   Reveal Solution Discussion Select an answer first
6. 25application · medium

   A retail company wants to implement an AI solution that predicts which customers are likely to churn in the next quarter, so the marketing team can target retention offers. Which type of AI should they deploy?

   • AComputer vision that analyzes customer demographics from photos
   • BPredictive AI that analyzes historical customer data to score churn likelihood
   • CReinforcement learning that optimizes email send times through trial and error
   • DGenerative AI that drafts personalized retention emails
   Reveal Solution Discussion Select an answer first