Plan, Create and Deploy a Microsoft Foundry Service AI-102 Practice Questions (Page 2)

Questions 6–10

1. 6application · medium

   A healthcare startup is building an AI solution to triage patient messages. They need to use a pre-built language model for sentiment analysis and a custom model for medical entity extraction, both accessed via a single endpoint. They want to minimize management overhead and avoid provisioning separate compute clusters. What should they do?

   • ACreate separate Azure AI resources for each model and configure a load balancer in front of them.
   • BUse Azure Machine Learning for both models and expose them via separate endpoints.
   • CCreate an Azure AI resource and deploy both models to the same endpoint using Azure AI Foundry.
   • DDeploy the pre-built model to Azure AI Foundry and the custom model to Azure Kubernetes Service, then use a custom gateway.
   Reveal Solution Discussion Select an answer first
2. 7application · medium

   A government agency is deploying an AI model to classify public records. They need to ensure that the model's decisions can be audited and that there is a clear chain of accountability. Which practice should they implement?

   • AUse a complex model that is difficult to interpret to prevent gaming.
   • BEnable detailed logging and monitoring for all model predictions and store logs for a defined retention period.
   • CAutomatically delete logs after 30 days to reduce storage costs.
   • DLimit access to the model's predictions to only a few administrators.
   Reveal Solution Discussion Select an answer first
4. 8application · medium

   A gaming company is using Azure AI Foundry to deploy a custom model for player behavior prediction. They want to access the model from their game server, which is hosted in a different Azure region. They need to configure the endpoint for the game server to use. What should they do?

   • AUse a private endpoint and connect the game server to the same virtual network.
   • BConfigure the game server to use the model's training data endpoint.
   • CCreate a new endpoint in the game server's region and redeploy the model there.
   • DUse the default endpoint provided by the Azure AI resource and ensure the game server has network access to it.
   Reveal Solution Discussion Select an answer first
5. 9foundation · easy

   Which Microsoft Responsible AI principle is most directly supported by establishing a clear process for reviewing and auditing AI model decisions?

   • AFairness
   • BPrivacy and security
   • CAccountability
   • DTransparency
   Reveal Solution Discussion Select an answer first
6. 10application · medium

   A company is deploying an AI system that provides medical diagnosis recommendations. They need to ensure that the system is transparent and that patients can understand how decisions are made. Which practice should they implement?

   • ALimit patient access to the system's recommendations to avoid confusion.
   • BUse a proprietary model to protect the intellectual property of the system.
   • CAutomatically delete all logs after each session to protect patient privacy.
   • DProvide patients with a clear explanation of the model's capabilities and limitations before use.
   Reveal Solution Discussion Select an answer first