Question 15
Domain 2: Fairness, Bias, and Societal ImpactA team is building a high-stakes AI system and wants to reduce bias throughout development and after deployment. Which approach best aligns with recommended bias-mitigation controls across the AI lifecycle?
Correct answer: D
Explanation
Bias mitigation should be addressed as a continuous lifecycle activity, with controls applied during data collection, preprocessing, model design, testing, and ongoing monitoring. — Controls across the AI lifecycle: propose controls during data collection, preprocessing, model design, testing, and monitoring.
Why each option is right or wrong
A. Focus on collecting representative data, then rely on deployment metrics without additional lifecycle controls.
Controls are proposed during data collection, preprocessing, model design, testing, and monitoring.
B. Apply bias checks during preprocessing and testing, because model design and monitoring have limited impact on fairness outcomes.
Controls are proposed during model design and monitoring as well as preprocessing and testing.
C. Concentrate bias mitigation in model design, since upstream data handling and downstream monitoring are outside the main control points.
Controls include data collection and monitoring, not only model design.
D. Implement bias-mitigation controls during data collection, preprocessing, model design, testing, and ongoing monitoring.
The source material explicitly identifies controls during data collection, preprocessing, model design, testing, and monitoring. Because the question asks for the approach that aligns with recommended controls across the AI lifecycle, the only complete answer is the one covering all five stages named in the source.