5.3 Explain AI Workflow Design and Monitoring Principles AI-TECHNICAL-PRACTITIONER Practice Questions (Page 2)

Questions 6–10

1. 6application · medium

   A retail company is building an AI workflow to forecast inventory demand. The workflow includes data ingestion from sales systems, preprocessing to handle missing values, model training, and deployment to a dashboard. The company wants to ensure that the workflow can be easily updated when new data sources are added. Which design principle is most important?

   • ADesign the data ingestion stage to be modular and configurable so new sources can be added without rewriting the entire pipeline.
   • BHard-code the data source connections into the preprocessing stage for efficiency.
   • CTrain the model on all available data at once, regardless of source.
   • DUse a single script that handles all stages to simplify maintenance.
   Reveal Solution Discussion Select an answer first
2. 7application · medium

   A logistics company is designing an AI workflow to optimize delivery routes. The workflow includes data ingestion from GPS devices, preprocessing to clean the data, training a routing model, and deploying it to a mobile app. The company wants to minimize the time between data collection and model update. Which design principle is most important?

   • ASeparate the training and deployment stages so they run on different schedules.
   • BAutomate the pipeline so that new data triggers preprocessing, training, and deployment without manual intervention.
   • CManually review and approve each model update before deployment.
   • DUse a large batch of historical data to train the model once and then deploy it permanently.
   Reveal Solution Discussion Select an answer first
4. 8foundation · easy

   What is the primary purpose of workflow orchestration in an AI pipeline?

   • ATo eliminate the need for model evaluation
   • BTo store all training data in a single database
   • CTo replace the need for data preprocessing
   • DTo automate the sequence and dependencies of workflow stages
   Reveal Solution Discussion Select an answer first
5. 9expert · hard

   A city government uses an AI model to allocate resources for public services. The model is monitored for accuracy, but there are concerns about bias against certain neighborhoods. The government wants to ensure that the model is fair and transparent. Which approach is most effective?

   • AUse a more complex model to reduce bias automatically.
   • BRely on the model's overall accuracy to ensure fairness.
   • CKeep the model's logic confidential to avoid public scrutiny.
   • DConduct a fairness audit and publish a report explaining the model's decisions and limitations.
   Reveal Solution Discussion Select an answer first
6. 10foundation · easy

   In an orchestrated AI workflow, what does managing dependencies between stages ensure?

   • AThat the model is deployed without any evaluation
   • BThat data is stored in a single location
   • CThat each stage runs only after its required inputs are ready
   • DThat all stages run simultaneously to save time
   Reveal Solution Discussion Select an answer first