AI Concerns, Ethical Principles, and Responsible AI CRAGE Practice Questions (Page 6)

Questions 26–30

1. 26application · medium

   A government agency is considering using an AI system to automate eligibility decisions for public benefits. The agency is concerned about public trust and wants to ensure the system is perceived as legitimate. Which combination of practices best supports responsible AI and public trust?

   • AUse a well-known commercial AI product to benefit from its brand reputation.
   • BKeep the system's logic confidential to prevent gaming and rely on internal quality checks.
   • CPublish an explanation of how the system works, establish an independent appeals process, and conduct regular audits for bias.
   • DDeploy the system only after achieving the highest possible accuracy on historical data.
   Reveal Solution Discussion Select an answer first
2. 27foundation · easy

   What is the primary purpose of responsible AI?

   • ATo maximize the speed of AI model training at any cost.
   • BTo guarantee that AI systems never make mistakes.
   • CTo replace human decision-making entirely with AI.
   • DTo ensure AI systems are developed and used in ways that are ethical, transparent, and accountable.
   Reveal Solution Discussion Select an answer first
4. 28application · medium

   An autonomous vehicle company is testing its AI in a new city. The AI has been trained mostly on data from suburban areas and now struggles with dense urban traffic patterns. Which technical concern of AI systems does this scenario primarily illustrate?

   • AData privacy violation, because the AI is collecting data from the new city without consent.
   • BLack of explainability, because the AI cannot explain why it struggles in urban areas.
   • CAlgorithmic bias, because the AI treats urban drivers differently than suburban drivers.
   • DGeneralization failure, where the model does not perform well on data that differs from its training distribution.
   Reveal Solution Discussion Select an answer first
5. 29application · medium

   A hospital deploys an AI triage system that prioritizes emergency patients. During validation, the data science team finds the model gives lower priority scores to patients from a specific postal code, which correlates with a lower-income neighborhood. The hospital's ethics committee asks for a mitigation before go-live. Which action best aligns with responsible AI principles?

   • ARetrain the model after applying a fairness-aware algorithm that adjusts for the identified bias while monitoring performance across subgroups.
   • BDeploy the model but add a disclaimer that the system is not intended for final diagnosis.
   • CDocument the bias in the model card and proceed with deployment because the model was trained on real-world data.
   • DRemove the postal code feature from the model and deploy it without further testing.
   Reveal Solution Discussion Select an answer first
6. 30application · medium

   A city government wants to deploy an AI system to allocate social services resources. The procurement team must choose between two vendors. Vendor A provides a detailed impact assessment, ongoing bias monitoring, and a clear appeals process. Vendor B offers a lower price and faster deployment but provides no documentation on how decisions are made. Which choice best reflects responsible AI procurement?

   • ASelect Vendor B because cost and speed are the primary considerations for public sector projects.
   • BSelect Vendor A because the system includes transparency, accountability, and ongoing oversight mechanisms.
   • CSelect Vendor A only if they agree to reduce their price to match Vendor B.
   • DSelect Vendor B and require them to add documentation after deployment.
   Reveal Solution Discussion Select an answer first