CAIC CAIC AI Governance & Compliance 2 — Questions and Answers
Question 1: What does 'data minimization' mean in the context of AI compliance?
- Compressing training datasets to reduce storage costs
- Collecting and using only the personal data necessary for the specified purpose (Correct answer)
- Removing outliers before model training
- Limiting the number of features in a dataset
Correct answer: Collecting and using only the personal data necessary for the specified purpose
Data minimization is a privacy principle requiring organizations to collect only the personal data strictly necessary for the AI system's stated purpose, reducing privacy risk.
Question 2: Which process involves continuously monitoring a deployed AI model for performance degradation over time?
- Model compression
- Model drift monitoring (Correct answer)
- Hyperparameter tuning
- Data labeling
Correct answer: Model drift monitoring
Model drift monitoring tracks whether a deployed model's accuracy or prediction quality degrades as real-world data distributions shift away from training data.
Question 3: In AI governance, what is a 'human-in-the-loop' (HITL) design?
- Using human-generated data for AI training
- Requiring human review or approval at key decision points in an automated AI process (Correct answer)
- Manual data pipeline orchestration
- Having humans label all model outputs
Correct answer: Requiring human review or approval at key decision points in an automated AI process
HITL design incorporates human judgment at critical stages of an AI workflow, ensuring oversight and the ability to override or correct automated decisions.
Question 4: What is 'right to explanation' as it applies to automated AI decisions affecting individuals in the US context?
- The right to request open-source access to model weights
- An individual's right to receive a meaningful explanation of how an automated decision that affects them was made (Correct answer)
- The vendor's right to explain pricing models
- A developer's right to document proprietary algorithms
Correct answer: An individual's right to receive a meaningful explanation of how an automated decision that affects them was made
The right to explanation gives individuals the ability to understand the logic behind automated decisions that significantly affect them, influenced by GDPR Article 22 and emerging US state laws.
Question 5: Which governance practice involves testing an AI system against adversarial inputs before deployment?
- A/B testing
- Red teaming (Correct answer)
- Blue-green deployment
- Canary releases
Correct answer: Red teaming
Red teaming for AI involves a team deliberately trying to elicit harmful, biased, or incorrect outputs from an AI system to identify vulnerabilities before public release.
Question 6: A company trains an AI model using customer purchase data without explicit consent. Which compliance violation is most directly applicable?
- Violation of SLA uptime requirements
- Violation of data privacy regulations such as CCPA or GDPR (Correct answer)
- Violation of software licensing agreements
- Violation of network security protocols
Correct answer: Violation of data privacy regulations such as CCPA or GDPR
Using personal data for AI training without consent or a lawful basis violates privacy regulations like GDPR and the California Consumer Privacy Act (CCPA).
CAIC CAIC AI Governance & Compliance 2
What does 'data minimization' mean in the context of AI compliance?