Ongoing outcome and prediction-quality metrics compared: To satisfy continuous-monitoring requirements, which

Tracking outcome and prediction-quality metrics against a baseline detects silent accuracy decay that uptime and volume signals miss.

The question

Preparing for AIGP? Take the free 5-min readiness quiz →

1. Server CPU and memory utilization over time, since resource-usage trends are the earliest indicator that a deployed model's predictions are beginning to degrade.

   Resource metrics reflect infrastructure load, not prediction quality, so they will not reveal rising default rates among approvals.
2. Ongoing outcome and prediction-quality metrics compared against a baseline, so gradual drops in real-world accuracy are detected despite no system errors. ✓

   Tracking outcome quality against a baseline surfaces silent accuracy decay that uptime and volume metrics cannot reveal.
3. The count of API requests received per hour, since a stable request volume confirms the model is still performing as expected for its user base.

   Request volume shows usage, not correctness; the scenario already notes volume is stable while accuracy silently degrades.
4. The application's error and exception logs, since a low error rate demonstrates the model continues to serve valid predictions to every incoming request.

   Error logs catch failures, not correct-but-wrong predictions; a model can return well-formed answers that are increasingly inaccurate.

How to remember it

Tracking outcome and prediction-quality metrics against a baseline detects silent accuracy decay that uptime and volume signals miss.

More Understanding How to Govern AI Development questions

• Lack of quality data: Which contributing factor most accurately explains the failure? →
• Specify the intended purpose: Which action most directly fulfills that objective? →
• Verifying and documenting a valid legal basis: Which step most directly addresses the requirement to confirm →
• All 135 Understanding How to Govern AI Development questions →