Question 160
At a multi-team data readiness review, an AI project leader faces this situation: A claims prediction team asks for every database export but has not defined the fields needed for the target outcome. The decision should distinguish missing evidence from confirmed data suitability. What acceptance evidence would best confirm successful action?
Explanation
Correct Answer
In a multi-team data readiness review, the decisive check is: Data owners approve a mapping from each required field to the planned training and evaluation use. That shows the planned action has worked: Define a feature-and-label specification with business meaning, expected type and source for each field. The governing reason is: Data requirements must specify meaningful predictors, labels, types and intended use rather than indiscriminate collection. A separate successful activity would not establish this specific requirement.
Incorrect Answers
Sensitive training attributes should be protected or removed when not necessary for the intended task. This alternative mechanism does not match the current fault.
A go/no-go decision needs adequate examples for critical outcomes and known performance uncertainty. Fixing that different mechanism would not address the current root cause: data requirements must specify meaningful predictors, labels, types and intended use rather than indiscriminate collection.
Subject-matter experts are needed to interpret data definitions and operational context correctly. Its verification is distinct from the reported condition: A claims prediction team asks for every database export but has not defined the fields needed for the target outcome. The decision should distinguish missing evidence from confirmed data suitability.