Conduct Risk Assessment(s) (e.g., Security, Safety, Ethics) PMI-CPMAI Practice Questions (Page 6)

Questions 26–30

1. 26foundation · easy

   What is the main objective of business continuity risk analysis for AI systems?

   • ATo assign blame for system failures
   • BTo eliminate all risks to the AI system
   • CTo identify and assess risks that could disrupt business operations
   • DTo ensure the AI system is always available
   Reveal Solution Discussion Select an answer first
2. 27application · medium

   A social media company uses an AI content moderation system to automatically remove hate speech and violent content. A news outlet reports that the system has been disproportionately removing content from minority groups, while allowing hate speech against these groups to remain. The story goes viral, and the company faces a public backlash. What is the most important immediate action for the company to take to manage its reputational risk?

   • AImmediately shut down the AI content moderation system and revert to human moderation for all content.
   • BBlame the news outlet for misrepresenting the system's performance and provide data showing the system is fair.
   • CQuietly work on fixing the algorithm and release an update without any public announcement.
   • DIssue a public statement acknowledging the issue, apologizing for the harm caused, and outlining a plan to investigate and fix the problem.
   Reveal Solution Discussion Select an answer first
4. 28foundation · easy

   Which of the following is an example of a risk mitigation strategy for an AI system that is vulnerable to adversarial attacks?

   • ARemoving all training data to prevent attacks
   • BIgnoring the risk because adversarial attacks are rare
   • CImplementing adversarial training to make the model more robust
   • DIncreasing the model's complexity to make it harder to attack
   Reveal Solution Discussion Select an answer first
5. 29foundation · easy

   What is the primary purpose of contingency planning in the context of AI-related risks?

   • ATo transfer AI risks to insurance providers
   • BTo prevent all AI failures from occurring
   • CTo document AI system performance metrics
   • DTo ensure business resilience by preparing responses to AI failures
   Reveal Solution Discussion Select an answer first
6. 30expert · hard

   A financial institution is planning to deploy an AI system to automate credit decisions. The system is expected to significantly reduce processing time and costs. However, an internal audit has identified a risk that the system may be biased against certain demographic groups. The institution's leadership is concerned about the potential reputational damage if this bias is discovered. They are evaluating different approaches to manage this risk. Which of the following approaches is the most effective in managing the reputational risk?

   • ADelay the deployment of the AI system until the bias is completely eliminated.
   • BDeploy the AI system and publicly deny any claims of bias, even if the audit findings are accurate.
   • CConduct a thorough ethical risk evaluation and implement a mitigation strategy to reduce bias, and be transparent about the system's limitations.
   • DDeploy the AI system but keep the details of its decision-making process confidential to avoid scrutiny.
   Reveal Solution Discussion Select an answer first