Task 1.1: Analyze Requirements and Design GenAI Solutions. AIP-C01 Practice Questions (Page 6)

Questions 26–30

1. 26foundation · easy

   An organization is choosing between two foundation models for a real-time translation service. Model A has higher accuracy but higher latency. Model B has lower latency but slightly lower accuracy. Which factor should be prioritized in the selection decision?

   • AThe number of developers on the team
   • BThe color scheme of the application's user interface
   • CThe application's user experience requirements for response time
   • DThe model's training data size
   Reveal Solution Discussion Select an answer first
2. 27foundation · easy

   A POC for a generative AI solution met all performance targets but exceeded the projected cost by 40%. What should the team do when evaluating the POC results?

   • AAutomatically cancel the project
   • BWeigh the cost overrun against the business value delivered by the solution
   • CIgnore the cost overrun because performance targets were met
   • DProceed to deployment without any changes
   Reveal Solution Discussion Select an answer first
4. 28application · medium

   An architect has designed a GenAI application that uses Amazon Bedrock, AWS Lambda, and Amazon DynamoDB. Before finalizing the design, the architect wants to identify potential improvements and best practices. Which AWS service or tool should the architect use?

   • AAWS Trusted Advisor
   • BAWS Config
   • CAmazon CloudWatch
   • DAWS Well-Architected Tool with the Generative AI Lens
   Reveal Solution Discussion Select an answer first
5. 29application · medium

   A media company needs to generate summaries of thousands of daily news articles. The summaries are stored in a data lake and are not needed in real-time. The company wants to minimize operational overhead and cost. Which integration pattern should the architect recommend?

   • AA streaming pipeline that processes articles as they are published and writes summaries to a database.
   • BA scheduled batch job that reads articles from the data lake, invokes the foundation model, and writes summaries back to the data lake.
   • CAn event-driven pattern where each new article triggers a serverless function that invokes the foundation model.
   • DA synchronous API invocation from the ingestion service to the foundation model for each article.
   Reveal Solution Discussion Select an answer first
6. 30expert · hard

   A company has completed a POC for a GenAI application. The POC was successful in terms of technical feasibility and user feedback. However, the estimated cost for full-scale deployment is significantly higher than the budget. The team has identified that the cost is driven by the high number of tokens processed. What is the most effective way to reduce token consumption?

   • AIncrease the max tokens setting to allow the model to generate longer, more complete responses.
   • BOptimize the prompts to be more concise and use techniques like few-shot examples to improve efficiency.
   • CSwitch to a larger, more capable model that requires fewer tokens to produce the same output.
   • DUse a batch processing pattern to reduce the number of API calls.
   Reveal Solution Discussion Select an answer first