Implement AI Solutions Responsibly AI-102 Practice Questions (Page 7)

Questions 31–35

1. 31application · medium

   A news portal uses Azure OpenAI to generate article summaries. They notice that the model sometimes produces text that incites violence when given certain controversial topics. The editorial team wants to automatically block such outputs before they are published. What should they implement?

   • AAdd a custom blocklist of controversial topic names.
   • BConfigure Azure Policy to deny access to the OpenAI endpoint for editors who work on controversial topics.
   • CUse prompt shields to block all prompts containing controversial topics.
   • DEnable the violence content filter in Azure AI Content Safety and set the severity threshold to 'Low'.
   Reveal Solution Discussion Select an answer first
2. 32foundation · easy

   In Azure AI Content Safety, what is the purpose of a blocklist?

   • ATo allow only certain IP addresses to access the AI service
   • BTo prevent the AI service from being used by unauthorized users
   • CTo block specific terms or phrases from being processed by the AI service
   • DTo automatically delete all content that contains sensitive data
   Reveal Solution Discussion Select an answer first
4. 33application · medium

   An online forum is using Azure AI Content Safety to moderate user posts. The forum has a list of banned words and phrases that must be blocked immediately, but the moderation team also wants to catch variations of these banned terms that might bypass the list. What should you configure?

   • AConfigure Azure Policy to block any post that contains a word from the banned list.
   • BSet the severity threshold to the lowest level so that all posts are flagged for manual review.
   • CUse the standard content filters only, since they are trained to detect all forms of banned content.
   • DAdd the banned terms to a custom blocklist and enable the 'match variations' option so that common misspellings and obfuscations are also detected.
   Reveal Solution Discussion Select an answer first
5. 34foundation · easy

   Which of the following is a key component of a responsible AI governance framework?

   • AEnsuring all AI models are deployed on-premises
   • BDefining clear roles and responsibilities for AI oversight
   • CChoosing the most expensive AI service
   • DGuaranteeing that AI models never make mistakes
   Reveal Solution Discussion Select an answer first
6. 35application · medium

   A dating app uses Azure AI Content Safety to moderate user profiles. They want to block profiles that contain explicit sexual content, but they also want to allow profiles that mention sexual orientation in a non-explicit way (e.g., 'I am gay'). What should they configure?

   • AAdd all sexual terms to a blocklist and set the action to 'Reject'.
   • BUse Azure AI Language to create a custom entity recognition model to detect sexual orientation.
   • CSet the sexual content severity threshold to 'Low' and rely on the AI to understand context.
   • DUse the Content Safety API with a custom blocklist of explicit terms and set the action to 'Review'.
   Reveal Solution Discussion Select an answer first