SnowPro Specialty: Gen AI Practice Questions

SnowPro Specialty: Gen AI (GES-C02) exam preparationAligned with the official GES-C02 Study Guide (last updated April 30, 2026)

SnowPro Specialty: Gen AI practice questions in English: 180 questions with explanations. The mock exam has 55 questions and 85 minutes.

SnowPro Specialty: Gen AI free question 1 / 10 · Domain 1: Snowflake for Gen AI Overview

A market research firm wants customers to ground their Cortex chatbots in the firm's licensed article archive as a Cortex Knowledge Extension. What does the firm build and share?

  1. A stage of article PDFs that consumers read with AI_PARSE_DOCUMENT.
  2. A model fine-tuned on the articles and logged in the Model Registry.
  3. A semantic view for Cortex Analyst over the article table, in a listing.
  4. A Cortex Search Service on the text table, shared in a listing.
Show the answer and explanation

Answer: D

Why this is correct
The page describes the flow in three steps. A provider uploads its text data into a table and creates a Cortex Search Service on that table, then shares the service on the Snowflake Marketplace or through organizational listings. A consumer builds a Cortex AI application, such as a chatbot, with Cortex AI Functions or the Cortex Agent API. When a prompt arrives, it is passed to the extension, which runs a semantic search and returns relevant knowledge to the application's LLM.

Why the other options are wrong
"A semantic view for Cortex Analyst over the article table, in a listing." … A semantic view describes structured data for text-to-SQL. A knowledge extension serves unstructured text by semantic search, which is not what a semantic view provides.
"A model fine-tuned on the articles and logged in the Model Registry." … A fine-tuned model is not what a provider shares here. The extension returns relevant passages at query time for retrieval-augmented generation, rather than baking the articles into model weights.
"A stage of article PDFs that consumers read with AI_PARSE_DOCUMENT." … Consumers do not receive raw files to parse. What is shared is a search service that already indexes the provider's text.

Why a search service
Knowledge extensions are a retrieval-augmented generation (RAG) building block: the consumer's model stays the same and the provider's content arrives as retrieved context. That is why the shared object is a Cortex Search Service and not a model or a dataset.

Go to the SnowPro Specialty: Gen AI practice app (10 free questions; purchase unlocks all questions)

See the full practice set