Prompt Engineering and Augmentation GENERATIVE-AI-ENGINEER-ASSOCIATE Practice Questions (Page 4)

Questions 16–20

1. 16application · medium

   A team is developing a chatbot for a bank. When a user asks 'What's my balance?', the baseline LLM responds with a generic explanation of what a balance is. The team wants the bot to actually retrieve the user's account balance. What is the most effective approach?

   • AAdd a constraint to 'be concise' to the baseline prompt.
   • BAccept the baseline response because it is informative.
   • CAugment the prompt with the user's account ID and a retrieved balance, then instruct the model to use that data in its response.
   • DUse a different prompt template that asks the model to guess the balance.
   Reveal Solution Discussion Select an answer first
2. 17application · medium

   A medical Q&A system receives: 'What are the side effects of Lisinopril for a patient with diabetes?' The team wants to retrieve relevant medical literature. Which term extraction is most appropriate for building the search query?

   • AExtract 'patient' and 'Q&A' as the key terms.
   • BExtract 'what', 'are', and 'the' as the key terms.
   • CExtract 'Lisinopril', 'side effects', and 'diabetes' as the key terms.
   • DExtract only 'Lisinopril' and ignore the rest.
   Reveal Solution Discussion Select an answer first
4. 18foundation · easy

   A user writes: "I need to troubleshoot my Spark cluster because the shuffle operations are slow." Which terms would be most important to extract as terms for augmentation?

   • ASpark, cluster, operations, slow
   • BI, need, to, because, the, are
   • CTroubleshoot, Spark, cluster, shuffle, slow
   • DTroubleshoot, slow, need, are
   Reveal Solution Discussion Select an answer first
5. 19expert · hard

   A developer is using an LLM to generate JSON from user input. The first prompt 'Convert this to JSON: {user input}' produces valid JSON but with extra fields the team doesn't want. The developer adds 'Only include the fields name and age.' The model still includes extra fields. What is the most effective next step?

   • AAdd a schema definition and an example of the exact JSON structure expected.
   • BRemove the instruction about fields and rely on the model's default behavior.
   • CUse a different LLM that is known to generate JSON better.
   • DIncrease the temperature to 0.2 to make the output more deterministic.
   Reveal Solution Discussion Select an answer first
6. 20foundation · easy

   What is a baseline LLM response?

   • AThe response the LLM gives when prompted without any additional context or specific instructions.
   • BThe response that is always factually correct.
   • CThe response from a smaller, less capable model.
   • DThe response generated after applying multiple prompt adjustments.
   Reveal Solution Discussion Select an answer first