Sustaining AI Transformation and Continuous Improvement CAIPM Practice Questions (Page 8)

Questions 36–40

1. 36expert · hard

   A European bank uses an AI system to assess creditworthiness. The system was developed in-house and is updated frequently. The bank's data protection officer (DPO) requires that the system comply with GDPR, including the right to explanation for automated decisions. The bank also wants to maintain the system's predictive accuracy. What is the best way to balance explainability and accuracy?

   • ARefuse to provide explanations, arguing that the model is proprietary.
   • BReplace the complex model with a simple, fully interpretable model, even if accuracy drops.
   • CUse explainable AI techniques, such as SHAP or LIME, to provide post-hoc explanations for each decision while retaining the high-accuracy model.
   • DOutsource the model to a third-party vendor that provides explanations as a service.
   Reveal Solution Discussion Select an answer first
2. 37expert · hard

   A professional services firm introduced an AI tool to automate report generation. The tool is technically sound, but adoption is low. The firm's partners are skeptical because they fear the tool will reduce the quality of client deliverables. The project sponsor wants to increase adoption quickly. What is the most effective strategy?

   • APilot the tool with a small group of partners and share the results with the broader team
   • BLaunch a mandatory training program and require all partners to use the tool
   • CDiscontinue the tool because partners are not adopting it
   • DOffer financial incentives to partners who use the tool
   Reveal Solution Discussion Select an answer first
4. 38application · medium

   A multinational corporation uses an AI system to screen job applicants. The legal department is concerned about potential bias and wants to ensure the system complies with equal employment opportunity laws. The company also wants to maintain the system's effectiveness. What governance practice should be implemented?

   • ARely on the AI vendor's assurance that the system is unbiased.
   • BConduct regular bias audits on the model's predictions and document the results for regulatory review.
   • CDisable the AI system to avoid any risk of discrimination.
   • DPublish the model's algorithm publicly to prove transparency.
   Reveal Solution Discussion Select an answer first
5. 39foundation · easy

   Which strategy is most effective for fostering an AI-driven culture in an organization?

   • ALimiting AI access to a few experts
   • BMandating AI usage without explanation
   • CFocusing only on technical implementation
   • DProviding continuous training and clear communication about AI's benefits and limitations
   Reveal Solution Discussion Select an answer first
6. 40expert · hard

   A retail company has an AI pricing model that adjusts prices in real time. The model's performance is measured by revenue and margin. Recently, the model increased revenue but decreased margin. The business team is concerned about the margin decline. The data science team believes the model is working as intended because revenue is up. What should the program manager do?

   • AAccept the revenue increase and ignore the margin decline
   • BAnalyze the trade-off between revenue and margin to determine if the model's behavior aligns with business goals
   • CChange the model's objective to maximize margin only
   • DImmediately revert to the previous pricing model
   Reveal Solution Discussion Select an answer first