CAIC Software Development Lifecycle 2

Question 1

Which SDLC phase is MOST critical for identifying AI model bias before a system goes to production?

Accepted Answer

Testing and validation

Answer

Testing and validation is the phase where bias, fairness, and accuracy issues are systematically identified and corrected before release.

Question 2

In an AI project using Scrum, what artifact BEST captures the evolving data requirements for model training?

Accepted Answer

Product backlog

Answer

The product backlog holds all prioritized work items, including data acquisition and preprocessing tasks, and is continuously refined.

Question 3

A team discovers that their AI model performs well in development but poorly in production. This is BEST described as:

Accepted Answer

Train-serve skew

Answer

Train-serve skew occurs when the data distribution or feature engineering in production differs from the training environment.

Question 4

Which practice BEST supports continuous integration in an AI development pipeline?

Accepted Answer

Automated testing of data pipelines and model performance

Answer

Automated testing of data pipelines and model metrics ensures that each code or data change is validated before merging.

Question 5

When defining the 'Definition of Done' for an AI feature, which criterion is MOST relevant?

Accepted Answer

Model accuracy meets the agreed threshold on a held-out test set

Answer

A measurable performance threshold on unseen data provides an objective, agreed criterion for completion of an AI feature.

(CAIC) Certified AI Consultant Practice Test