Privacy, Ethical, Trust and Safety Controls AAISM Practice Questions (Page 1)

Questions 1–5

1. 1foundation · easy

   Which factor is most directly associated with building trust in an AI system by ensuring that there is a clear process for holding the organization responsible for the system's outcomes?

   • AEfficiency
   • BAccountability
   • CUsability
   • DRobustness
   Reveal Solution Discussion Select an answer first
2. 2expert · hard

   A credit union uses an AI system to approve loan applications. The system is a deep neural network that is highly accurate but is a 'black box'. The credit union's regulators require that they be able to explain the specific reasons for each loan denial. The credit union wants to maintain the model's accuracy while meeting the regulatory requirement. What is the best approach?

   • AUse a post-hoc explainability technique, such as SHAP, to generate feature attributions for each denial.
   • BReplace the neural network with a logistic regression model that is inherently interpretable.
   • CHire a data scientist to manually review each denial and provide an explanation.
   • DProvide a generic explanation, such as 'the model determined you do not meet our criteria'.
   Reveal Solution Discussion Select an answer first
4. 3application · medium

   A healthcare organization is deploying an AI system that processes patient data to predict readmission risks. The system is hosted in a public cloud region that complies with local data residency laws. The organization wants to ensure that patient data is protected and that the AI model does not inadvertently expose personal information in its outputs. Which combination of controls should the organization implement?

   • AStore all patient data on-premises to avoid any cloud-related privacy risks.
   • BRely solely on the cloud provider's encryption and access controls to protect patient data.
   • CImplement data minimization, pseudonymization, and output filtering to remove personal data from model responses.
   • DAnonymize all data before training and rely on the model's inherent inability to memorize personal data.
   Reveal Solution Discussion Select an answer first
5. 4foundation · easy

   Which data governance practice is most directly responsible for defining who within an organization is allowed to access and modify personal data used in AI systems?

   • AData anonymization
   • BData lineage
   • CData access control
   • DData backup
   Reveal Solution Discussion Select an answer first
6. 5foundation · easy

   Which ethical principle requires that AI systems be designed and operated in a way that avoids causing unjustified harm to individuals or society?

   • AJustice
   • BAutonomy
   • CNon-maleficence
   • DBeneficence
   Reveal Solution Discussion Select an answer first