Policy Design CRAGE Practice Questions (Page 1)

Questions 1–5

1. 1expert · hard

   A multinational corporation is developing an AI recruitment tool. The policy team must engage stakeholders across different regions, including employee representatives, job applicants, and local regulators. There are conflicting views: some regions want strict transparency, while others prioritize efficiency. The policy must be globally consistent yet locally adaptable. Which engagement strategy best achieves this?

   • AEngage only with regulators to ensure legal compliance, and then impose the policy on all regions.
   • BHold a single global town hall meeting and incorporate feedback into one uniform policy.
   • CLet each regional office develop its own policy independently, with no global coordination.
   • DCreate a global policy framework with core principles, and establish regional advisory committees to adapt the policy to local contexts.
   Reveal Solution Discussion Select an answer first
2. 2expert · hard

   A global technology company is designing an AI governance policy for a facial recognition system used in public spaces. The policy must respect human rights, including privacy and freedom of assembly, while also meeting the company's business objectives. The company operates in multiple countries with varying legal frameworks. Which policy approach best addresses the ethical and legal complexities?

   • ARely on local legal teams to determine the ethical acceptability of the system in each location.
   • BAdopt a human rights impact assessment as a mandatory step before deploying the system in any new location.
   • CLimit the system's use to locations where it is legally permitted, without additional ethical review.
   • DDeploy the system in all locations and address human rights concerns only if complaints are raised.
   Reveal Solution Discussion Select an answer first
4. 3expert · hard

   An AI company is designing a policy for a deepfake detection tool. The tool can be used for positive purposes, such as identifying misinformation, but also has potential for misuse, such as surveillance. The policy must consider human rights and societal impact. Which policy element is most critical?

   • ALimit the policy to legal compliance, such as data protection laws, without addressing ethical use cases.
   • BFocus solely on the technical accuracy of the tool, assuming that accurate detection will naturally lead to positive outcomes.
   • CInclude a clear statement of intended use and prohibited uses, along with a mechanism for ongoing human rights impact assessments.
   • DRely on customer agreements to ensure that buyers use the tool responsibly.
   Reveal Solution Discussion Select an answer first
5. 4application · medium

   A financial institution has implemented an AI governance policy that requires model risk assessments before deployment. The policy is being ignored by some business units because they see it as a bottleneck. Which enforcement mechanism would best improve compliance without completely halting innovation?

   • AIncrease the penalties for non-compliance, such as suspending projects that skip the assessment.
   • BIntroduce a streamlined risk assessment process with clear timelines and a fast-track for low-risk models, while maintaining full review for high-risk models.
   • CRequire all models to undergo the same full risk assessment, regardless of their potential impact.
   • DEliminate the risk assessment requirement for models that are not customer-facing.
   Reveal Solution Discussion Select an answer first
6. 5application · medium

   A financial services firm has adopted an AI governance policy that requires bias testing for all customer-facing models. The policy is approved, but the data science team is not consistently running the required tests before deployment. Which enforcement mechanism would most effectively ensure compliance with the policy?

   • AIntegrate the bias testing requirement into the CI/CD pipeline so models cannot be deployed without passing the tests.
   • BSend a monthly reminder email to the data science team about the policy requirement.
   • CAsk the data science team to self-report their compliance status at the end of each quarter.
   • DAdd the requirement to the team's annual performance review without changing the deployment process.
   Reveal Solution Discussion Select an answer first