Objective 1.1 Identify and Describe Common Terminology or Concepts Important to Data-Driven Technology Ethics CERTIFIED-ETHICAL-EMERGING-TECHNOLOGIST Practice Questions (Page 4)

Questions 16–20

1. 16application · medium

   A US-based company collects personal data from users in California and in the European Union. The company wants to use this data for targeted advertising. Which legal obligation must the company address?

   • AOnly GDPR applies because it is stricter than CCPA
   • BOnly CCPA applies because the company is US-based
   • CBoth CCPA and GDPR may apply, depending on the residency of each data subject
   • DNeither law applies because the data is used for commercial purposes
   Reveal Solution Discussion Select an answer first
2. 17expert · hard

   A credit card company develops a fraud detection model. The model is evaluated on a dataset where only 1% of transactions are fraudulent. The model achieves 99% accuracy. However, the company is concerned about the model's ability to catch fraud. Which evaluation metric is most appropriate to assess the model's performance in this context, and why?

   • AF1 score, because it balances precision and recall
   • BAccuracy, because it is the most intuitive metric for overall performance
   • CPrecision, because it measures the proportion of predicted fraud cases that are actually fraud
   • DRecall, because it measures the proportion of actual fraud cases correctly identified
   Reveal Solution Discussion Select an answer first
4. 18application · medium

   A university develops a natural language processing model to screen student essays for plagiarism. The model is trained on essays from previous years. The university finds that the model flags essays from non-native English speakers as plagiarized more often than essays from native speakers. Which type of bias is most likely at play, and what metric would best quantify the disparity?

   • ASocietal bias; precision
   • BConfirmation bias; recall
   • CData bias; false positive rate by language background
   • DAlgorithmic bias; overall accuracy
   Reveal Solution Discussion Select an answer first
5. 19expert · hard

   A city government deploys a computer vision system in public spaces to detect and alert on unattended bags for security purposes. The system uses facial recognition to identify individuals near the bags. Privacy advocates raise concerns. The city argues the system serves public safety. Which ethical decision-making approach best balances the competing concerns?

   • ARely solely on the legal opinion that the system is lawful under current surveillance laws
   • BConduct a proportionality assessment weighing the security benefit against the privacy intrusion, and consider less intrusive alternatives
   • CDiscontinue the system entirely because any privacy intrusion is unacceptable
   • DApply a deontological approach: always prioritize the duty to protect public safety, regardless of privacy impact
   Reveal Solution Discussion Select an answer first
6. 20expert · hard

   A recruitment firm develops an AI system to screen job applicants. The system uses a neural network trained on resumes of employees hired in the past 10 years. The firm discovers that the system ranks male candidates higher than female candidates with similar qualifications. The firm wants to fix this while maintaining the system's predictive performance. Which approach is most appropriate?

   • AManually adjust the model's outputs to promote more female candidates
   • BUse a fairness-aware training technique that optimizes for both accuracy and demographic parity
   • CRemove gender from the training data and retrain the model
   • DAbandon the AI system and use manual resume screening
   Reveal Solution Discussion Select an answer first