Shadow, champion-challenger and A/B deployment
Ways to introduce a new model safely.
In shadow mode the new model runs alongside the live one without affecting decisions, so outputs can be compared. In champion-challenger, the current model (champion) is compared with a candidate (challenger) on part of the traffic. A/B testing splits users between versions to measure outcomes.
On the exam
Shadow deployment lets a model be tested on real data with no impact on customers.
Related terms
- Change management for AIControlling changes to models, data, prompts, thresholds and settings.
- Model validation and testingChecking that a model is fit for its intended use before and after deployment.
AAIA Test yourself on this with timed AAIA practice questions in the AAIA Exam Simulation Hub, or read the free AAIA exam guide.
Test yourself with today’s question of the day.
