Databricks Certified Generative AI Engineer Associate — Question 51

A Generative AI Engineer is building a production-ready LLM system which replies directly to customers. The solution makes use of the Foundation Model API via provisioned throughput. They are concerned that the LLM could potentially respond in a toxic or otherwise unsafe way. They also wish to perform this with the least amount of effort.

Which approach will do this?

Answer options

• A. Ask users to report unsafe responses
• B. Host Llama Guard on Foundation Model API and use it to detect unsafe responses.
• C. Add some LLM calls to their chain to detect unsafe content before returning text
• D. Add a regex expression on inputs and outputs to detect unsafe responses.

Correct answer: B

Explanation

The correct answer is B because hosting Llama Guard on the Foundation Model API provides a dedicated solution for detecting unsafe responses with minimal additional effort. Option A relies on user reporting, which can be inconsistent and slow. Option C requires additional LLM calls, increasing complexity, while Option D involves regex, which may not effectively capture all unsafe responses.