A team is preparing a production rollout and needs to avoid a design mistake. The team is focused on fairness evaluation. Which recommendation is most appropriate?
Select an answer to reveal the explanation.
Short Explanation and Infographic
A nutrition label that only lists total calories — not sugar, fat, or sodium — is technically accurate and completely misleading. Option A is the full nutrition panel for AI: evaluate outcomes for protected and relevant groups wherever fairness risk exists. Aggregate accuracy is the total-calories-only approach. Skipping sensitive-data checks is avoiding the hard columns. Majority-group-only testing omits the populations most likely to be harmed. All groups on the label — that is option A!
Full explanation below image
Full Explanation
The correct answer is A. Evaluate outcomes for protected or relevant groups where fairness risk exists. aligns directly with the 4.0 Configure Evaluation and Monitoring exam objective and reflects how IBM watsonx.governance operationalizes this practice in enterprise environments. IBM watsonx.governance is designed to provide the tooling, workflows, and evidence collection mechanisms that make AI trustworthy and auditable from intake through retirement. Choosing this option ensures the implementation is both defensible to regulators and maintainable by governance teams across the full AI lifecycle. Option B is incorrect because it does not address the core governance requirement described in the scenario and would leave an important control gap in production. Option C is incorrect because it eliminates a required control and leaves the governance program without a critical safeguard. Option D is incorrect because it narrows the scope to a single artifact or metric and misses the full governance requirement. On scenario-based exam questions, the correct answer is consistently the one that operationalizes the IBM governance practice with structured controls rather than approximating it with shortcuts, manual workarounds, or oversimplified rules.