Responsible Practices and Governance Standards AIE Practice Questions (Page 7)

Questions 31–35

1. 31application · medium

   A city government is planning to deploy an AI system to prioritize housing assistance applications. The system will use historical data that may reflect past biases. The project team includes data scientists and city officials, but no community representatives. Which approach best addresses the ethical risk of bias while incorporating stakeholder input?

   • AHire an external consultant to build the model without sharing data or involving stakeholders.
   • BUse the historical data as-is and rely on the city council to review the system annually.
   • CExclude all demographic variables from the model to guarantee fairness.
   • DConduct a bias audit on the historical data and form a stakeholder advisory group that includes community members and advocacy groups.
   Reveal Solution Discussion Select an answer first
2. 32application · medium

   A financial institution deploys an AI system that approves or rejects loan applications. Regulators require the institution to demonstrate that the system's decisions can be reviewed and challenged. The institution has a team of compliance officers who are not data scientists. Which mechanism best supports accountability in this scenario?

   • AAutomatically log all decisions but restrict access to the model development team only.
   • BReplace the AI system with a rule-based system to eliminate the need for explainability.
   • CGrant compliance officers direct access to the raw training dataset for manual inspection.
   • DProvide compliance officers with a dashboard that shows the key factors influencing each decision in plain language.
   Reveal Solution Discussion Select an answer first
4. 33foundation · easy

   Which of the following is an accountability mechanism commonly used in AI systems?

   • AMaintaining detailed documentation of AI model development and decisions.
   • BDeleting logs after each model update to save storage.
   • CUsing proprietary algorithms without any external review.
   • DAutomating all decisions without human oversight.
   Reveal Solution Discussion Select an answer first
5. 34application · medium

   An e-commerce company has deployed a recommendation system. After a few months, they notice that the system's recommendations are becoming less diverse and may be reinforcing users' existing preferences, potentially leading to filter bubbles. The company wants to maintain ethical standards and improve the system. What should they do first?

   • ADisable the recommendation system and rely on manual curation.
   • BSet up continuous monitoring of recommendation diversity and user satisfaction metrics, and schedule periodic model reviews.
   • CIgnore the issue because the system is still generating sales.
   • DImmediately retrain the model with more data to improve accuracy.
   Reveal Solution Discussion Select an answer first
6. 35foundation · easy

   Which of the following is an example of incorporating stakeholder input into AI governance decision-making?

   • AConducting public consultations to gather feedback on an AI system's design.
   • BKeeping all AI system details confidential to avoid external influence.
   • CMaking decisions solely based on the preferences of the largest shareholder.
   • DLimiting discussions to the internal development team only.
   Reveal Solution Discussion Select an answer first