Operationalize AI Solution
Free PMI Certified Professional in Managing AI practice — 6 questions on Operationalize AI Solution, with explanations. No sign-up. Full 12-question mixed test →
A fraud-detection model has been deployed to production for six months. The model's precision has dropped from 0.91 to 0.68, but investigation shows the input feature distributions remain statistically unchanged from training data. The relationship between transaction patterns and fraud outcomes has shifted due to new fraud tactics. Which phenomenon is occurring, and what governance action is MOST appropriate?
When input feature distributions are stable but the underlying relationship between inputs and the target outcome has changed, this is concept drift, not data drift. The correct governance response is retraining with fresh, representative labeled data capturing the new pattern.
An organization is deploying a new AI-powered loan-approval model to replace a legacy rules engine. Leadership wants to validate the model's real-world behavior on live production traffic without exposing customers to any risk if the model performs poorly. Which deployment strategy BEST satisfies this requirement?
Shadow deployment runs the new model alongside the incumbent system on real production inputs, generating predictions for comparison without those predictions ever driving actual customer-facing decisions, eliminating customer risk entirely.
A CPMAI practitioner is finalizing the model governance plan for a production churn-prediction model. Which trigger condition should be explicitly configured to initiate an unscheduled retraining cycle, rather than waiting for the next scheduled retraining interval?
Threshold-based alerting tied to monitored performance metrics is the correct configuration for triggering unscheduled, event-driven retraining, ensuring the model is refreshed as soon as measurable degradation is detected rather than waiting on a fixed calendar.
During post-deployment monitoring, an AI-driven inventory-forecasting model begins producing recommendations that lead to significant stockouts across multiple warehouses within 48 hours of a new version release. The contingency plan for this solution defines an incident response procedure. What should the team do FIRST, according to CPMAI contingency planning best practices?
Contingency plans for AI incidents should define an immediate mitigation step—typically rollback to a known-stable version—to halt business harm first, before root-cause analysis, retraining, or reporting activities proceed.
As a project transitions from the project delivery team to operational support (BAU), the CPMAI transition plan must define clear ownership. Six weeks after go-live, a model's prediction latency begins exceeding SLA thresholds. No one on the support team acted for two days because it was unclear who owned infrastructure-level performance monitoring versus who owned model-quality monitoring. What gap does this scenario BEST illustrate?
The delay resulted from ambiguity about who was responsible for acting on the alert, not a lack of detection. This is a classic gap in the transition plan's role definitions (RACI), which CPMAI stresses must be explicit when handing off from project team to operational support.
An organization's AI governance framework requires quarterly disaster recovery (DR) testing for all production AI solutions classified as business-critical. During a scheduled DR test, the team simulates a full outage of the model-serving endpoint. Which outcome would indicate the contingency plan is functioning as intended?
A successful DR test validates that predefined failover mechanisms (backup rules engine, cached results, redundant endpoints, etc.) activate and restore business continuity within the recovery time objective, which is the actual purpose of regular contingency plan testing.
Ready for the real thing?
The full course: two full-length practice tests, video lessons for every exam domain, hands-on labs and detailed explanations.
$109.99 $34.99 with code FREETEST33 — valid through September 14.
Get my $34.99 deal →