Translate Business Use Case Goals into a Description of the Desired Inputs and Outputs for the AI Pipeline GENERATIVE-AI-ENGINEER-ASSOCIATE Practice Questions (Page 4)

Questions 16–20

1. 16application · medium

   A bank wants to build a pipeline to approve small business loans. The loan officers need a decision recommendation within 5 minutes of application submission. They have application data, credit scores, and financial statements. What is the most important alignment consideration between inputs and outputs?

   • AThe output must be in a printed format for physical filing.
   • BThe output must be a detailed 50-page report for each application.
   • CThe pipeline must use only historical data from the past 10 years.
   • DThe input data must be available in a format that allows the pipeline to generate a decision within the 5-minute window.
   Reveal Solution Discussion Select an answer first
2. 17application · medium

   A financial services firm wants to build an AI pipeline to detect fraudulent transactions. They have transaction data with timestamps, amounts, merchant categories, and customer profiles. The compliance team requires that the pipeline output an alert for every transaction with a fraud probability above 0.8, and they need to review alerts within 10 minutes of the transaction. What is the most important constraint to consider when specifying the pipeline's inputs and outputs?

   • AThe pipeline must output alerts in a PDF report format.
   • BThe pipeline must store all transaction data for at least 7 years for regulatory compliance.
   • CThe pipeline must process transactions in near real-time to meet the 10-minute review window.
   • DThe pipeline must use only structured data inputs.
   Reveal Solution Discussion Select an answer first
4. 18application · medium

   An insurance company wants to build a pipeline to automate claim triage. The goal is to route claims to the appropriate adjuster based on complexity. They have claim forms, photos, and policy documents. The adjusters need a triage decision within 1 hour of claim submission. What input specification is most appropriate?

   • AHistorical claim data from the past year, without current submissions.
   • BOnly the claim forms, without photos or policy documents.
   • CReal-time streaming of claim photos as they are uploaded.
   • DBatch of claim forms, photos, and policy documents submitted within the last hour, processed every hour.
   Reveal Solution Discussion Select an answer first
5. 19foundation · easy

   A marketing team wants to generate personalized email subject lines. The pipeline must process customer data in near real-time. Which input-output alignment is most appropriate?

   • AInput: customer purchase history and browsing behavior; Output: a subject line for each customer, generated within seconds of a trigger event.
   • BInput: customer age and location; Output: a weekly email newsletter.
   • CInput: all customer emails; Output: a list of subject lines for all customers, generated daily.
   • DInput: customer feedback surveys; Output: a subject line for each customer, generated within seconds.
   Reveal Solution Discussion Select an answer first
6. 20expert · hard

   A global e-commerce company wants to build a pipeline to recommend products to users. The goal is to increase cross-selling. They have user purchase history, browsing behavior, and product inventory. The marketing team wants real-time recommendations on the website, while the data science team wants to use a complex model that requires batch processing. The company has a limited budget for infrastructure. What is the most appropriate approach to align inputs and outputs?

   • AUse batch processing only and update recommendations daily.
   • BUse real-time streaming to run the complex model on every user action.
   • CUse a hybrid approach: precompute recommendations in batch and serve them in real-time from a fast lookup store.
   • DUse a simple model that runs in real-time, ignoring the complex model.
   Reveal Solution Discussion Select an answer first