Which practice best reflects sound lifecycle governance for AI models already in production?
Select an answer to reveal the explanation.
Short Explanation
You watch for the changes that matter—data, purpose, version, or risk—and reassess when they happen. Don't wait for a complaint or a calendar date to tell you something's off.
Full Explanation
The sound practice is continuous monitoring for material changes in data, purpose, version, or risk, with reassessment triggered whenever one occurs. This lets governance respond promptly rather than depending on an outdated approval or an arbitrary schedule. The practice of relying on the initial approval indefinitely is incorrect because production status reflects conditions at deployment time, not guaranteed accuracy of that assessment indefinitely afterward. The practice of reassessing only when a customer complaint is received is incorrect because relying solely on complaints means problems are caught only after users are already affected. The practice of reassessing on a fixed annual schedule regardless of what has or has not changed in between is incorrect because a fixed schedule can miss urgent changes that occur between review dates, or waste effort reviewing when nothing material has changed. For the exam, favour change-triggered reassessment over purely time-based or complaint-driven review.