A team preparing a production rollout wants to avoid gaps in its governance evidence. Why is waiting until release to start tracking a model risky?
Select an answer to reveal the explanation.
Short Explanation
If you only start the paper trail at go-live, everything before it is gone. You lose the intake, development and validation decisions that show why the model was built that way. That leaves no evidence trail for later review.
Full Explanation
The correct answer is the option stating that waiting until release leaves no record of intake, development and validation decisions, so those choices cannot later be reviewed or justified. Intake, development and validation set assumptions, data sources and test results, which reviewers, auditors and incident investigators need to see. The option about failing automated build steps is incorrect because tracking is governance, not a build-pipeline mechanism. The option about preventing production registration is incorrect because a model can still be registered late; the issue is missing history. The option about doubling compute cost is incorrect because tracking does not change compute requirements. For the exam, treat late tracking as an evidence gap, not a technical blocker.