3.1 Explain Responsible AI Principles (fairness, Transparency, Accountability, Bias Mitigation, Safety) AI-TECHNICAL-PRACTITIONER Practice Questions (Page 3)

Questions 11–15

1. 11expert · hard · select all that apply

   A healthcare AI system is being developed to predict patient readmission risk. The team discovers that the model performs poorly for patients with rare conditions. Which actions would be appropriate to mitigate this bias and improve safety? (Select all that apply.)

   • AUse data augmentation techniques to create synthetic examples of rare conditions.
   • BCollect more data from patients with rare conditions to improve model performance for these groups.
   • CImplement a fallback mechanism that flags low-confidence predictions for human review.
   • DRemove rare conditions from the model's scope to avoid errors.
   • EIncrease the model's complexity to better capture rare condition patterns.
   Check answers Discussion Select an answer first
2. 12expert · hard

   A social media platform uses an AI system to moderate content. The system is found to disproportionately flag content from certain minority groups. The platform wants to reduce this bias without compromising safety (e.g., allowing harmful content). What is the best strategy?

   • AAdjust the model's decision threshold for minority groups to reduce false positives, while monitoring for harmful content.
   • BLower the overall sensitivity of the model to reduce all flags, including harmful content.
   • CRemove the minority group labels from the training data to make the model group-agnostic.
   • DUse a completely different model architecture to avoid bias.
   Reveal Solution Discussion Select an answer first
4. 13foundation · easy

   What is the difference between explainability and interpretability in AI?

   • AExplainability is about the data used, while interpretability is about the algorithm
   • BExplainability is about understanding the model's internal logic, while interpretability is about explaining individual decisions
   • CExplainability refers to the ability to explain individual decisions, while interpretability refers to understanding the overall model behavior
   • DThey are the same concept and can be used interchangeably
   Reveal Solution Discussion Select an answer first
5. 14application · medium

   An autonomous vehicle company is testing its AI system in a simulated environment. The system performs well in normal traffic but fails in rare, unexpected situations. What is the most important safety consideration?

   • AEnsure the system can handle edge cases and unexpected scenarios safely.
   • BRely on the system's high accuracy in simulations to ensure safety.
   • CFocus on improving the system's performance in normal traffic conditions.
   • DIncrease the complexity of the simulation to match real-world conditions exactly.
   Reveal Solution Discussion Select an answer first
6. 15foundation · easy

   What does accountability in AI involve?

   • AEnsuring that AI systems are always correct and never fail
   • BKeeping the AI development process secret to avoid external scrutiny
   • CMaking AI systems fully autonomous without human intervention
   • DAssigning clear roles and responsibilities for AI outcomes and establishing oversight mechanisms
   Reveal Solution Discussion Select an answer first