Transparency CRAGE Practice Questions (Page 8)

Questions 36–40

1. 36application · medium

   A medical device company develops an AI system that assists radiologists in detecting tumors. The company must provide transparency to regulatory authorities. Which combination of mechanisms is most appropriate?

   • ADetailed technical documentation, audit logs, and clear labeling of the system's intended use and limitations
   • BA user-friendly mobile app for patients to view their own scans
   • CAn internal wiki accessible only to the development team
   • DA press release announcing the system's capabilities
   Reveal Solution Discussion Select an answer first
2. 37application · medium

   A university uses an AI system to screen student applications. The admissions office wants to communicate the system's role to prospective students. Which message best aligns with transparency principles?

   • AOur AI system is highly accurate and will ensure a fair review of your application.
   • BThe AI system's decisions are final and cannot be appealed.
   • CWe use an AI system to assist in reviewing applications, but final decisions are made by our admissions committee.
   • DWe do not use AI in admissions, to protect the integrity of the process.
   Reveal Solution Discussion Select an answer first
4. 38expert · hard

   A government agency is deploying an AI system to assist with benefits eligibility determinations. The agency must balance the public's right to understand decisions with the need to protect the system's security and prevent gaming. Which transparency approach best reconciles these competing requirements?

   • AProvide individualized explanations to applicants while keeping the model's internal logic confidential
   • BPublish the full model code and training data to maximize transparency
   • CKeep all information about the AI system confidential to prevent gaming
   • DPublish a general description of the system but refuse to explain individual decisions
   Reveal Solution Discussion Select an answer first
5. 39foundation · easy

   Why is explainability important for AI transparency, even when a model is highly interpretable?

   • ABecause explainability is required by law, while interpretability is not
   • BBecause interpretable models are always less accurate, so explanations are needed to justify their use
   • CBecause stakeholders often need clear reasons for specific decisions, not just an understanding of the model's general logic
   • DBecause interpretability only applies to unsupervised learning models
   Reveal Solution Discussion Select an answer first
6. 40application · medium

   An insurance company uses an AI system to set premiums. The company wants to enable internal auditors to review why specific premium decisions were made. Which transparency mechanism is most effective for this purpose?

   • APublishing a general privacy policy on the company website
   • BComprehensive logging of inputs, outputs, and model versions for each decision
   • CUsing a highly interpretable model that requires no logging
   • DProviding customers with a chatbot to ask questions about their premiums
   Reveal Solution Discussion Select an answer first