Building Privacy, Trust, and Safety in AI Systems CRAGE Practice Questions (Page 2)

Questions 6–10

1. 6application · medium

   An e-commerce company is deploying an AI recommendation system. A risk assessment has identified that the system could potentially recommend inappropriate products to minors. The company wants to mitigate this risk. Which mitigation strategy is most effective?

   • ADisable the recommendation system for all users under the age of 18.
   • BAdd a disclaimer to the website stating that the recommendations are not tailored to minors.
   • CImplement an age-verification mechanism and filter the recommendation algorithm's output to exclude products that are not age-appropriate.
   • DConduct a DPIA to further analyze the risk, but do not implement any changes yet.
   Reveal Solution Discussion Select an answer first
2. 7application · medium

   A fitness app company is adding an AI feature that provides personalized health recommendations based on user activity data. The company must comply with data protection regulations that grant users the right to access, rectify, and erase their data. The product team is designing the data management interface. Which feature is essential to respect user data rights?

   • AAllowing users to download their data in a proprietary format that is difficult to read.
   • BRequiring users to submit a written request by mail to access their data.
   • CProviding users with a dashboard where they can view, correct, and delete their activity data.
   • DAutomatically deleting user data after 30 days without any option for users to access it.
   Reveal Solution Discussion Select an answer first
4. 8application · medium

   A bank uses an AI system to approve or deny loan applications. Customers who are denied loans often ask why. The bank wants to improve customer trust and comply with regulations that require meaningful information about the logic involved in automated decisions. Which approach best addresses the need for explainability and transparency?

   • ASending a generic notice that the decision was made by an automated system.
   • BPublishing a general description of the AI model's architecture on the bank's website.
   • COffering customers a toll-free number to call for a general overview of the loan approval process.
   • DProviding each denied customer with a specific explanation of the key factors that influenced the decision.
   Reveal Solution Discussion Select an answer first
5. 9application · medium

   An insurance company uses an AI system to determine premiums for auto insurance. Regulators require that customers be informed about the logic involved in automated decisions. The company wants to ensure that customers understand why they received a particular premium. Which approach best satisfies the requirement for explainability?

   • ASending a generic statement that premiums are based on a proprietary algorithm.
   • BOffering customers a chatbot that answers general questions about insurance.
   • CPublishing a general description of the AI model's methodology on the company's website.
   • DProviding each customer with a personalized notice that lists the key factors (e.g., driving record, age, location) that influenced their premium.
   Reveal Solution Discussion Select an answer first
6. 10application · medium

   A financial institution is deploying an AI system that approves or denies loan applications. The compliance team is concerned about regulatory scrutiny and customer complaints. The system has shown high accuracy in testing, but the institution wants to build user trust and ensure accountability. Which combination of measures should the institution prioritize to address these concerns?

   • ARely solely on automated decisions without human intervention to ensure consistency and speed across all applications.
   • BProvide customers with a clear explanation of the key factors that influenced the decision and establish a process for human review upon request.
   • CKeep the model's decision logic confidential to protect the institution's competitive advantage and only publish aggregate approval rates.
   • DIncrease the model's complexity to improve accuracy, even if the decision-making process becomes less interpretable to customers.
   Reveal Solution Discussion Select an answer first