Transparency, Interpretability and Explainability CT-AI Practice Questions (Page 1)

Questions 1–5

1. 1application · medium

   An insurance company is deploying an AI system to set premiums. The company operates in a jurisdiction where regulations mandate that individuals have the right to obtain an explanation for decisions that significantly affect them. The company's current model is a deep neural network. What is the most critical requirement the company must meet to comply with this regulation?

   • AEnsure the system can provide a meaningful, understandable explanation for each individual's premium decision.
   • BProvide a general statement that the system uses a complex algorithm to set premiums.
   • CReplace the deep neural network with a simple linear regression model to ensure full interpretability.
   • DPublish the model's source code and training data on a public website.
   Reveal Solution Discussion Select an answer first
2. 2foundation · easy

   How is 'interpretability' best distinguished from 'explainability'?

   • AInterpretability focuses on the data; explainability focuses on the code
   • BInterpretability is only relevant for deep learning models; explainability is for all other models
   • CInterpretability is a regulatory requirement; explainability is a technical feature
   • DInterpretability is about the model itself; explainability is about providing post-hoc explanations for decisions
   Reveal Solution Discussion Select an answer first
4. 3application · medium

   A team of data scientists has built a highly accurate deep learning model for image classification. They are now trying to understand which parts of an image the model focuses on when making a classification. They use a technique that highlights the pixels most influential in the model's decision. This technique is best described as an example of:

   • AGlobal interpretability, as it describes the model's overall behavior.
   • BLocal explainability, as it explains a specific prediction.
   • CTransparency, as it makes the model's inner workings visible.
   • DGlobal explainability, as it explains the model's logic in general.
   Reveal Solution Discussion Select an answer first
5. 4application · medium

   A university is deploying an AI system to screen student applications. The admissions committee wants to understand the model's overall logic to ensure it aligns with the university's values. A rejected applicant wants to know what specific factors led to their rejection. The university's legal team wants to ensure the system is not discriminating against protected groups. Which set of needs and corresponding AI quality characteristics is correctly matched?

   • AAdmissions committee: interpretability (local); rejected applicant: explainability (global); legal team: transparency.
   • BAdmissions committee: interpretability (global); rejected applicant: explainability (local); legal team: transparency.
   • CAdmissions committee: explainability (local); rejected applicant: interpretability (global); legal team: transparency.
   • DAdmissions committee: transparency; rejected applicant: interpretability (global); legal team: explainability (local).
   Reveal Solution Discussion Select an answer first
6. 5foundation · easy

   In which scenario is explainability most critical for an AI-based system?

   • AA system that predicts the weather for the next week
   • BA system that recommends movies based on user preferences
   • CA system that approves or rejects loan applications
   • DA system that sorts emails into spam and non-spam folders
   Reveal Solution Discussion Select an answer first