AI Model Deployment Strategies: Canary and A/B Testing
An MLOps team uses a CI/CD pipeline to automate model retraining. The pipeline triggers on new labeled data, runs feature engineering, retrains the model, evaluates against a holdout set, and deploys if metrics exceed thresholds. Recently, a retrained model passed validation but caused a 5% accuracy drop in production. Which improvement best prevents this?
⚠ Common exam trap
A common misconception is that more data or larger validation sets always improve model reliability. However, the trap here is that distribution drift between training/validation and live production is the real cause of accuracy drops, which only online evaluation methods like canary deployment can detect.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Implement canary deployment with shadow scoring to compare with current model
Canary deployment with shadow scoring allows the new model to serve predictions to a small subset of traffic while comparing its outputs against the current production model in real time, without affecting all users. This catches subtle data drift or concept drift that a static holdout set may miss, preventing the 5% accuracy drop from reaching full production.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✓
Implement canary deployment with shadow scoring to compare with current model
Why this is correct
Canary deployment with shadow scoring routes live traffic to the retrained model in parallel, comparing its predictions against the incumbent before full promotion. This catches the production accuracy drop that holdout validation missed, because the stem's failure arose from a validation-to-production gap, not from flawed training metrics.
- ✗
Require manual approval before deployment
Why it's wrong here
Manual approval adds a human gate but no detection mechanism; the reviewer sees the same passing holdout metrics, so the production drop still ships. It is tempting as a governance control, and would be correct where regulatory sign-off is mandated, but it does not address the distribution shift causing the accuracy loss.
- ✗
Use the entire production dataset for validation instead of a holdout set
Why it's wrong here
Validating on the full production dataset leaks training data into evaluation, inflating metrics and hiding the very drift that caused the drop. It is tempting because more data feels more representative, and would be correct for final post-deployment monitoring, not for pre-deployment validation against unseen data.
- ✗
Increase the amount of training data used in each retraining cycle
Why it's wrong here
More training data does not correct a validation set that no longer matches production conditions, so the accuracy drop recurs. It is tempting because data volume usually improves generalisation, and would be correct when the model is underfitting, but here the failure is distribution shift between holdout and live traffic.
About these practice questions
Courseiva writes every AI0-001 question from scratch — 962 in total, each with an explanation and a wrong-answer breakdown. None are copied from real exams or dumps. Learn why practice questions differ from exam dumps →
JA
Written by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
This AI0-001 practice question is part of Courseiva's free CompTIA certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the AI0-001 exam.