nerdexam
Databricks

GENERATIVE-AI-ENGINEER-ASSOCIATE · Question #29

A Generative AI Engineer at an automotive company would like to build a question-answering chatbot for customers to inquire about their vehicles. They have a database containing various documents of…

The correct answer is A. Invite users to submit long, rather than concise, questions. A RAG-based vehicle Q&A chatbot requires three core components: an embedding model (to convert documents and queries into vector representations), a vector database (to store and retrieve those embeddings via semantic search), and a response-generating LLM (to synthesize…

Retrieval Augmented Generation (RAG) Architecture

Question

A Generative AI Engineer at an automotive company would like to build a question-answering chatbot for customers to inquire about their vehicles. They have a database containing various documents of different vehicle makes, their hardware parts, and common maintenance information. Which of the following components will NOT be useful in building such a chatbot?

Options

  • AInvite users to submit long, rather than concise, questions
  • BResponse-generating LLM
  • CEmbedding model
  • DVector database

How the community answered

(34 responses)
  • A
    94% (32)
  • B
    3% (1)
  • C
    3% (1)

Explanation

A RAG-based vehicle Q&A chatbot requires three core components: an embedding model (to convert documents and queries into vector representations), a vector database (to store and retrieve those embeddings via semantic search), and a response-generating LLM (to synthesize answers from retrieved context). Encouraging users to submit long questions is actively counterproductive - verbose, unfocused queries degrade semantic search quality because the embedding becomes diluted across multiple topics. Best practice is to guide users toward concise, specific questions to maximize retrieval precision. Therefore, option A is the component that is NOT useful.

Topics

#RAG architecture#Chatbot components#Embedding models#Vector databases

Community Discussion

No community discussion yet for this question.

Full GENERATIVE-AI-ENGINEER-ASSOCIATE Practice