Transparency, Interpretability and Explainability CT-AI Practice Questions (Page 6)

Questions 26–28

1. 26application · medium

   A healthcare provider is using an AI system to recommend treatment plans. A doctor is reviewing a specific patient's case and wants to know why the AI recommended a particular drug. The hospital's quality assurance team wants to verify that the AI's recommendations are consistent with medical guidelines across all patients. Which interpretability techniques should be used?

   • AUse a global surrogate model, like a decision tree, for both the doctor and the quality assurance team.
   • BUse a global explanation technique like feature importance for the doctor, and a local technique like LIME for the quality assurance team.
   • CUse the same SHAP summary plot for both the doctor and the quality assurance team.
   • DUse a local explanation technique like LIME for the doctor, and a global technique like feature importance for the quality assurance team.
   Reveal Solution Discussion Select an answer first
2. 27foundation · easy

   How do regulators typically view transparency and explainability in AI systems?

   • AAs concerns only for the developers, not for the regulators
   • BAs essential requirements for ensuring accountability and enabling oversight
   • CAs technical details that are irrelevant to compliance
   • DAs optional features that are nice to have but not required
   Reveal Solution Discussion Select an answer first
3. 28foundation · easy

   Which stakeholder group is most likely to require explanations for individual decisions made by an AI system?

   • ASoftware developers who are building the system
   • BData engineers who are preparing the training data
   • CEnd users who are directly affected by the system's decisions
   • DCloud infrastructure providers who host the system
   Reveal Solution Discussion Select an answer first