AI0-001 AI Implementation and Operations Practice Question

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✓

  Scheduled monitoring of key performance metrics such as AUC, precision, and recall against recent labeled outcomes

  Why this is correct

  Tracking performance metrics on recent labeled data detects degradation as customer behavior evolves, giving the firm evidence that the deployed model still performs as validated. Because labels for credit outcomes arrive with a delay, scheduling these evaluations ensures drift is caught before it materially harms decisions, which directly supports continuous assurance.

• ✓

  Regular fairness audits comparing approval rates and error rates across protected groups

  Why this is correct

  Continuous fairness assurance requires periodically measuring outcomes such as approval rates and false positive rates across protected groups. Fairness can degrade even when overall accuracy holds steady, so scheduled audits comparing group-level metrics are essential to confirm the model continues to meet the firm's ethical and regulatory obligations over time.

• ✗

  Retraining the model on the full historical dataset every night regardless of any detected change

  Why it's wrong here

  Blind nightly retraining on all historical data is expensive and can entrench stale patterns rather than adapt to new behavior. Without monitoring evidence or validation, scheduled retraining may also silently introduce regressions. Continuous assurance comes from measurement and targeted action, not from unconditionally rebuilding the model every night.

• ✗

  Storing all inference requests and responses in immutable logs for later forensic analysis

  Why it's wrong here

  Immutable logging supports auditability and incident investigation, which is valuable, but logging alone does not measure whether performance or fairness has degraded. Without analysis of those logs against outcomes and group metrics, the firm would have records but no assurance, so logging is a supporting practice rather than one of the two best answers.

• ✗

  Increasing the model's parameter count to improve its capacity to fit customer data

  Why it's wrong here

  Adding parameters changes model capacity but does nothing to verify ongoing performance or fairness. A larger model can even overfit and worsen generalization. The compliance objective is observability of the live system, which is achieved through monitoring and audits rather than through architectural changes to the model itself.