Question 380
At a data engineering team preparing AI inputs, an AI project leader faces this situation: Multiple departments provide the same customer attribute with inconsistent definitions and no known owner. The team must reconcile source quality, permission and readiness evidence. Which safeguard should guide the team during implementation?
Explanation
Correct Answer
In a data engineering team preparing AI inputs, the key precaution is: Do not overwrite a source record merely to make separate departments' data agree. The planned action is: Engage the designated data steward to agree a canonical definition and conflict-resolution policy. This caution matters because: Data stewards govern shared definitions, source precedence and quality resolution across organizational boundaries. The team should still verify: Downstream features use one approved attribute definition and record unresolved exceptions.
Incorrect Answers
Source discovery should identify authoritative records and duplication across internal applications. The appropriate correction for the reported case is: Engage the designated data steward to agree a canonical definition and conflict-resolution policy.
A model's data location and collection interface must match latency and sampling requirements. Fixing that different mechanism would not address the current root cause: data stewards govern shared definitions, source precedence and quality resolution across organizational boundaries.
Reliable collection must make replay and checkpoint commits consistent with target writes. Its verification is distinct from the reported condition: Multiple departments provide the same customer attribute with inconsistent definitions and no known owner. The team must reconcile source quality, permission and readiness evidence.