GENERATIVE-AI-ENGINEER-ASSOCIATE · Question #49
A Generative AI Engineer is building a system which will answer questions on latest stock news articles. Which will NOT help with ensuring the outputs are relevant to financial news?
The correct answer is B. Increase the compute to improve processing speed of questions to allow greater relevancy. Increasing compute power improves processing speed but has no direct effect on the relevance or accuracy of outputs to a specific domain like financial news.
Question
A Generative AI Engineer is building a system which will answer questions on latest stock news articles. Which will NOT help with ensuring the outputs are relevant to financial news?
Options
- AImplement a comprehensive guardrail framework that includes policies for content filters tailored
- BIncrease the compute to improve processing speed of questions to allow greater relevancy
- CImplement a profanity filter to screen out offensive language.
- DIncorporate manual reviews to correct any problematic outputs prior to sending to the users
How the community answered
(38 responses)- B92% (35)
- C3% (1)
- D5% (2)
Why each option
Increasing compute power improves processing speed but has no direct effect on the relevance or accuracy of outputs to a specific domain like financial news.
A guardrail framework with domain-specific content filters directly restricts outputs to relevant financial topics, improving relevance.
Relevance of LLM outputs to financial news is governed by factors like prompt engineering, retrieval strategies, guardrails, and content filtering - not raw compute speed. Increasing compute only reduces latency; it does not constrain or guide the model to stay on-topic with financial content. A faster irrelevant answer is still irrelevant.
A profanity filter helps ensure outputs meet content standards, which is part of a safe and appropriate response policy for financial news.
Manual reviews catch and correct problematic or off-topic outputs before they reach users, directly improving relevance and safety.
Concept tested: LLM output relevance control mechanisms
Source: https://docs.databricks.com/en/generative-ai/guardrails.html
Topics
Community Discussion
No community discussion yet for this question.