A Generative AI Engineer is creating a customer support bot that should respond differently to an end user based on the sentiment in their initial message. For example, if the end user's message was…

Question

Options

• AUse an encoder-only LLM model to both detect sentiment and generate replies based upon the
• BImplement a RAG architecture for how to respond to users depending on detected sentiment.
• CUse linear regression model to classify sentiment and feed the result to a system prompt for the
• DCreate a chain which first uses an LLM to classify sentiment, then changes system prompt for the

• A
  2% (1)
• B
  14% (6)
• C
  5% (2)
• D
  80% (35)

Explanation

The best practice for this use case is a two-step LLM chain: first, use an LLM to classify the sentiment of the user's message (e.g., angry, neutral, happy), then dynamically adjust the system prompt for the response-generating LLM based on that classification (e.g., injecting de-escalation instructions when anger is detected). Option A is incorrect because encoder-only models (like BERT) are designed for classification/understanding tasks only and cannot generate text responses. Option B (RAG for sentiment routing) adds unnecessary complexity and is not the right tool for this pattern. Option C uses linear regression, which is an outdated, lower-accuracy approach compared to using an LLM for nuanced sentiment classification. Option D correctly separates concerns into a clean, maintainable chain.

No community discussion yet for this question.