Ensure Responsible AI Use CY0-001 Practice Questions (Page 2)

Questions 6–10

1. 6application · hard

   A marketing company wants to use a cloud-based AI service to analyze customer purchase history and send personalized offers. The data includes names, email addresses, and purchase amounts. The company's legal team requires that the AI service provider cannot read the raw customer data. Which privacy-preserving technique should the company use?

   • AAnonymize the data by removing names and email addresses before sending it to the AI service.
   • BSend the data to the AI service provider over a TLS-encrypted connection.
   • CStore the data in a private cloud and only send aggregate statistics to the AI service.
   • DUse a homomorphic encryption scheme that allows computation on encrypted data.
   Reveal Solution Discussion Select an answer first
2. 7foundation · easy

   What is the main goal of privacy-preserving techniques in AI, such as differential privacy?

   • ATo protect individual data while still allowing useful analysis
   • BTo reduce the size of the training dataset
   • CTo improve the accuracy of the AI model
   • DTo make the AI model easier to interpret
   Reveal Solution Discussion Select an answer first
4. 8application · medium

   A retail company uses an AI system to determine which customers should receive a discount coupon. The system is a gradient boosting model. A customer complains that they did not receive a coupon and wants to know why. The company wants to provide a transparent explanation. Which information is most useful to share with the customer?

   • AA list of the top three factors that most influenced the decision, such as purchase frequency and average order value.
   • BA statement that the decision was made by an AI and is confidential.
   • CThe customer's rank among all other customers in the model's scoring.
   • DThe exact mathematical formula of the gradient boosting model.
   Reveal Solution Discussion Select an answer first
5. 9foundation · easy

   Which ethical principle is most directly concerned with ensuring that AI systems do not cause disproportionate harm to individuals or groups, even when pursuing beneficial outcomes?

   • AJustice
   • BAutonomy
   • CBeneficence
   • DNon-maleficence
   Reveal Solution Discussion Select an answer first
6. 10foundation · easy

   What is the primary purpose of testing an AI model for bias before deployment?

   • ATo ensure the model performs consistently across different demographic groups
   • BTo improve the model's overall accuracy on the training data
   • CTo verify that the model meets all regulatory compliance requirements
   • DTo reduce the computational resources required for inference
   Reveal Solution Discussion Select an answer first