Task 1.6: Implement Prompt Engineering Strategies and Governance for FM Interactions. AIP-C01 Practice Questions (Page 2)

Questions 6–10

1. 6foundation · easy

   A prompt system needs to pass the output of one prompt as input to the next prompt in a fixed order. Which design pattern does this describe?

   • ASequential chaining
   • BPost-processing steps
   • CReusable components
   • DConditional branching
   Reveal Solution Discussion Select an answer first
2. 7foundation · easy

   A developer is creating a prompt template for a customer-support chatbot. The template must define the chatbot's persona, set boundaries for topics it may discuss, and enforce a consistent response format. Which combination of elements should the developer include in the instruction framework?

   • AA few-shot examples, a max token limit, and a stop sequence
   • BA system prompt, a user prompt, and a temperature setting
   • CA role definition, guardrails, and a template configuration
   • DA model identifier, an API endpoint, and an authentication token
   Reveal Solution Discussion Select an answer first
4. 8application · medium

   A software company is using Amazon Bedrock to generate code documentation. The initial prompt 'Write documentation for this code' produces inconsistent results: sometimes the documentation is too verbose, sometimes too brief, and it often misses edge cases. The team wants to iteratively improve the prompt. Which approach is the MOST systematic way to enhance the prompt's performance?

   • AUse a different, larger foundation model that is known to be better at code generation, and keep the prompt the same.
   • BAdd a chain-of-thought instruction to the prompt: 'Think step by step about the code before writing the documentation.'
   • CCreate a test set of 20 diverse code samples. Define clear output specifications (e.g., 'Include a summary, a list of parameters, and a list of edge cases. Keep the total length under 200 words.'). Run the prompt on the test set, evaluate the outputs against the specifications, and iterate on the prompt based on the failures.
   • DManually tweak the prompt based on gut feeling after reviewing a few outputs, and test it on a new piece of code each time.
   Reveal Solution Discussion Select an answer first
5. 9foundation · easy

   An interactive AI application needs to orchestrate a sequence of steps: analyze user sentiment, retrieve context, and then call a foundation model. Which AWS service is designed to coordinate these steps as a workflow?

   • AAWS Step Functions
   • BAmazon Comprehend
   • CAmazon SageMaker
   • DAmazon DynamoDB
   Reveal Solution Discussion Select an answer first
6. 10expert · hard

   A financial analysis firm is building a system that generates investment reports. The workflow is: 1) Analyze the user's risk profile, 2) Retrieve relevant market data, 3) Generate a draft report, and 4) Perform a compliance check for regulatory language. The team has built this as a single, complex prompt. The problem is that the model sometimes fails the compliance check, and when it does, the entire report is unusable. The team wants to improve the reliability of the system. Which approach is the MOST effective?

   • AUse a larger, more powerful foundation model that is less likely to generate non-compliant text.
   • BKeep the single prompt but add a stronger instruction: 'You MUST pass the compliance check. Do not generate any text that could be considered non-compliant.'
   • CBreak the workflow into a sequential chain of separate prompts using Step Functions. If the compliance check fails, the system automatically retries the generation step with a modified prompt that includes the specific compliance failure reason.
   • DAdd a post-processing step in the Lambda function that uses string matching to detect and remove any non-compliant phrases from the generated report.
   Reveal Solution Discussion Select an answer first