Treat the disparity as a material risk and remediate it: Which response best fits risk management during the
A large group-level disparity found in testing is a material risk that must be remediated and revalidated before release.
The question
During testing, a fraud model shows strong overall accuracy but flags one ethnic group at triple the false-positive rate of others. The business wants to ship on schedule. Which response best fits risk management during the training-and-testing phase?
Preparing for AIGP? Take the free 5-min readiness quiz →
- Ship on schedule and add the disparity to the product backlog, since overall model accuracy already meets the agreed project target.Plausible under deadline pressure, but shipping a known disparate harm defers an unacceptable risk into production.
- Lower the global decision threshold for everyone, which cuts the total number of flags raised but leaves the relative gap between groups intact.Plausible as a quick tuning move, but a uniform threshold change does not close the disparity between groups.
- Treat the disparity as a material risk and remediate it before release, then revalidate to confirm the mitigation actually reduced the harm. ✓Correct: a large group-level disparity is a material risk that must be mitigated and re-tested before deployment, not deferred.
- Exclude the affected group from the evaluation set so the reported metrics no longer show the disparity before the release review.Plausible as a way to clear the dashboard, but hiding the metric conceals the harm rather than managing it.
The trap
Letting strong overall accuracy or deadline pressure justify deferring, masking, or cosmetically patching a group-level disparity. How to remember it
A large group-level disparity found in testing is a material risk that must be remediated and revalidated before release.
How many of these would you get right?
One of 1581 AIGP questions on Certsqill. Take a free five-minute check and see your score per domain — not one number, but which section to open tonight.
Test your AIGP readiness — freeMore Understanding How to Govern AI Development questions
- Bias testing across protected groups together: Beyond validation of predictive performance, which combination →
- Records of the datasets used: Which documentation most directly lets the auditor validate results and confirm →
- To document the model's intended use: What is the primary purpose of a model card at release? →
- All 426 Understanding How to Govern AI Development questions →
Part of the Certsqill AIGP question bank · Understanding How to Govern AI Development ·
Every answer, right and wrong, comes with its own explanation.
