Databricks Generative AI Engineer Associate Question 123
Single answerYou are building a customer support chatbot using a large language model (LLM) in Databricks. A user submits the following input: 'I need help with my recent order, specifically the refund process.' To generate a more accurate response, you want to augment the LLM prompt with additional context extracted from the input. Which of the following approaches would best ensure the LLM understands the user's intent and generates the most helpful response?
- A
Extract specific keywords like 'order' and 'refund' and include them in the prompt to focus the LLM on the topic.
- B
Provide the LLM with a generic prompt such as 'Answer the question based on the input provided.'
- C
Analyze the user's input for intent and context, then include a summary like 'The user wants assistance with processing a refund for a recent order.' in the prompt.
- D
Include the entire user input verbatim in the prompt without any additional processing.
Show answer and explanation
Correct answer: C
Explanation
The most effective approach to augmenting a prompt is to analyze the user's input for key fields, terms, and intent, then summarize it into a clear and concise context. This ensures the LLM focuses on the user's specific needs and generates a helpful response, avoiding generic or ambiguous outputs.
- A. Incorrect.
While extracting keywords like 'order' and 'refund' can help identify important terms, it lacks sufficient detail about the user's intent, which might lead to a less accurate response.
- B. Incorrect.
Using a generic prompt like 'Answer the question based on the input provided' does not provide enough context for the LLM to generate a specific or targeted response.
- C. Correct.
Summarizing the user's intent and context, such as specifying that the user wants help with a refund for a recent order, provides clear guidance to the LLM, improving the accuracy and relevance of the response.
- D. Incorrect.
Including the entire user input verbatim without processing could lead to ambiguity, as the LLM might not fully understand the user's intent or focus on the key aspects of the input.