Assess representativeness and label quality across: What evidence should it prioritize?
The evidence gap concerns whether data adequately represent affected locations and whether multilingual labels support reliable prioritization.
The question
Assume EU AI Act high-risk duties apply to a travel-support system that prioritizes passenger cases. Historical training data cover major airports well but contain sparse records from regional airports and inconsistent labels across languages. The team has a data inventory but no analysis of representativeness or label quality. What evidence should it prioritize?
Preparing for AIGP? Take the free 5-min readiness quiz →
- Add more computing-capacity records to the inventory.Computing-capacity records describe infrastructure, not whether regional coverage and multilingual labels are suitable for the system’s intended purpose.
- Compare only overall accuracy against the previous travel-support version.Aggregate accuracy can conceal regional and language-specific weaknesses, leaving the stated representativeness and label-quality gaps unresolved.
- Publish the vendor’s general privacy policy alongside the inventory.A privacy policy may describe processing practices but does not establish data quality, coverage, or label consistency for this system.
- Assess representativeness and label quality across affected groups and locations. ✓The observed geographic imbalance and inconsistent labels require evidence about relevance, representativeness, quality, and foreseeable performance effects.
The trap
When the stem names coverage and label problems, select evidence addressing those properties directly rather than aggregate performance. How to remember it
The evidence gap concerns whether data adequately represent affected locations and whether multilingual labels support reliable prioritization.
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 Laws, Standards and Frameworks Apply to AI questions
- Document mitigations and post-deployment verification: Which documentation action best addresses the evidence →
- Name trained overseers and authorize pause: What additional control is most necessary? →
- Begin post-market monitoring: Which requirement does the evidence most directly trigger? →
- All 394 Understanding How Laws, Standards and Frameworks Apply to AI questions →
Part of the Certsqill AIGP question bank · Understanding How Laws, Standards and Frameworks Apply to AI ·
Every answer, right and wrong, comes with its own explanation.
