Question 31
Domain 4: Assembling and Deploying ApplicationsWhy might a team use SQL AI Functions to prototype an offline extraction or classification baseline?
Correct answer: A
Explanation
SQL AI Functions let a team test LLM behavior directly against existing tables, so they can "evaluate LLM behavior close to existing data" before investing in a full app. That makes them useful for prototyping an offline extraction or classification baseline because the team can measure model output on real data with minimal setup.
Why each option is right or wrong
A. They let the team evaluate LLM behavior close to existing data without building a full application first
SQL AI Functions are designed to run model inference directly in SQL over existing tables, so a team can score or label real rows without first engineering an application layer or orchestration service. In practice, that means they can benchmark extraction/classification behavior on in-database data immediately, using the same source records they would later operationalize, which is why this is a fast way to prototype an offline baseline.
B. They replace all need for model serving
They help with prototyping and evaluation, but production model serving may still be needed.
C. They only work for vector search indexing
They can support broader LLM tasks like extraction and classification, not just vector indexing.
D. They are intended only for collecting user feedback
They are for running AI logic on data, not only gathering feedback from users.