Question 420
At fictional Ironvale Learning Services, a Snowflake data and AI engineering team is conducting a multi-tenant knowledge assistant improvement before any automated output can change customer or financial records. The separately composed technical incident is: A procurement department records AI-generated recommendations but not the source versions, model settings, human approvals or overrides that determined final purchase actions. Assume only these documented facts and use safe, authorized production-change procedures; exact current Snowflake feature syntax and region or account availability require current review. Which repeatable assurance or control is most relevant to preventing this situation?
Explanation
Writer-proposed answer: D (UNVERIFIED)
Preliminary individual option explanations — NOT expert validated
Correct Answer
Define minimum decision-record fields, retention and access policies, and regularly test end-to-end traceability using sample cases. Accountable AI governance separates automated advice from the final authorized decision and preserves enough evidence to investigate errors.
Incorrect Answers
This response concerns Redaction quality on complex identifiers, which is different from the stated failure. A procurement department records AI-generated recommendations but not the source versions, model settings, human approvals or overrides that determined final purchase actions.
Gen AI runaway cost and chargeback is insufficient to demonstrate this result: An independent auditor can reproduce the decision pathway from authorized records and distinguish model suggestions from the human's approved outcome.
Cortex model access allowlist governance is not the cause of the reported failure: A procurement department records AI-generated recommendations but not the source versions, model settings, human approvals or overrides that determined final purchase actions.