Implement AI Solutions Responsibly AI-102 Practice Questions (Page 4)

Questions 16–20

1. 16application · medium

   A company is deploying an Azure OpenAI-based assistant that helps employees draft emails. The security team is concerned about prompt injection attacks where a user might trick the assistant into generating harmful content or revealing internal information. What should you implement to mitigate this risk?

   • AConfigure the model's temperature to 0 so that it always produces the same safe response.
   • BEnable prompt shields in Azure AI Content Safety to detect and block prompt injection attempts before they reach the model.
   • CUse Azure Policy to restrict the assistant to only respond to emails that contain the company's domain.
   • DAdd a system message that instructs the model to ignore any attempts to change its behavior.
   Reveal Solution Discussion Select an answer first
2. 17expert · hard

   A financial services firm is deploying an Azure OpenAI-based assistant that helps advisors generate client emails. The firm has a strict policy that the assistant must not produce content that could be considered financial advice unless it is directly quoted from approved documents. They also want to prevent the assistant from being tricked into revealing confidential client information. Which two Azure AI capabilities should they combine?

   • APrompt shields, and a custom blocklist of financial terms.
   • BAzure AI Search with semantic ranking, and a private endpoint for the OpenAI service.
   • CGroundedness detection with approved documents as the source, and prompt shields.
   • DContent filters with a high severity threshold, and custom blocklists of client names.
   Reveal Solution Discussion Select an answer first
4. 18foundation · easy

   What is the primary purpose of a responsible AI governance framework?

   • ATo speed up the deployment of AI models
   • BTo increase the accuracy of AI models
   • CTo establish policies and practices that ensure AI systems are developed and used ethically
   • DTo reduce the cost of AI development
   Reveal Solution Discussion Select an answer first
5. 19application · medium

   A healthcare organization is deploying an Azure OpenAI-based chatbot that provides medical information to patients. The compliance team requires that the organization can demonstrate how the AI system makes decisions and that there is a process for human review of flagged interactions. What should you configure to meet these requirements?

   • ADeploy the chatbot to a separate virtual network and restrict access to the compliance team only.
   • BEnable Azure AI Studio's responsible AI insights, which include debugging and monitoring tools, and set up a human review workflow for flagged conversations.
   • CUse Azure Policy to enforce that all chatbot responses are reviewed by a clinician before being sent to patients.
   • DConfigure Azure Monitor to collect all chatbot logs and set up alerts for any conversation that includes medical terms.
   Reveal Solution Discussion Select an answer first
6. 20application · medium

   A government agency is deploying an AI system to assist with public service requests. The agency must ensure that the AI system is transparent and accountable to the public. What should be part of the governance framework?

   • AEstablish a review board that oversees AI deployments, publish transparency reports, and use responsible AI insights to monitor system behavior.
   • BKeep all AI system details confidential to protect intellectual property.
   • CUse a simple rule-based system instead of AI to avoid accountability issues.
   • DAllow the AI system to operate without monitoring to reduce costs.
   Reveal Solution Discussion Select an answer first