Implement AI Solutions Responsibly AI-102 Practice Questions (Page 2)

Questions 6–10

1. 6expert · hard

   A video-sharing platform uses Azure AI Content Safety to moderate user-uploaded videos. The platform has a policy to block videos containing hate speech, but the moderation team is concerned about videos that use sarcasm or irony to convey hate. The team also wants to block specific hate symbols that are not in the standard filter. What should you do?

   • ADisable all content filters and rely on user reports to identify hate speech.
   • BSet the severity threshold to the lowest level to flag all videos for manual review.
   • CUse the hate and fairness content filter for general hate speech, add a custom blocklist for specific hate symbols, and enable prompt shields to detect manipulative captions or comments that might indicate sarcastic hate.
   • DUse only the violence content filter, since it is the most comprehensive.
   Reveal Solution Discussion Select an answer first
2. 7foundation · easy

   What is the primary purpose of 'responsible AI insights' in an Azure AI solution?

   • ATo monitor and evaluate AI systems for fairness, reliability, and safety
   • BTo reduce the cost of AI service usage
   • CTo increase the speed of AI model training
   • DTo automatically fix any bugs in the AI model
   Reveal Solution Discussion Select an answer first
4. 8expert · hard

   A multinational bank is deploying an AI system to approve loan applications. The bank must comply with regulations that require fairness and transparency, but the AI team is under pressure to deploy quickly. The bank also needs to block certain types of applications that are considered high-risk. What is the best approach to balance speed and compliance?

   • AImplement a governance framework that includes fairness testing, bias mitigation, and ongoing monitoring using responsible AI insights, while using custom blocklists to flag high-risk applications.
   • BSkip fairness testing to save time, since the model is based on a well-known algorithm.
   • CUse a simple rule-based system to avoid AI compliance requirements.
   • DDeploy the AI system immediately and address compliance issues after launch.
   Reveal Solution Discussion Select an answer first
5. 9expert · hard

   A multinational bank is deploying an Azure OpenAI-based virtual assistant for internal employee HR queries. The legal team requires that all AI-generated responses be traceable to specific policy documents, and the security team requires that any attempt to extract unrelated internal data be blocked. The bank also needs to demonstrate to regulators that they have a process for reviewing and updating AI safety policies. Which combination of Azure AI capabilities should the bank implement?

   • AAzure AI Search with semantic ranking, a private endpoint for the OpenAI service, and a weekly automated report of all user queries.
   • BContent filters set to the strictest level, a custom blocklist of competitor names, and a data retention policy of 90 days.
   • CGroundedness detection with source citation, prompt shields, and a documented governance review cycle.
   • DCustom blocklists for all internal document names, content filters for hate speech, and a one-time policy document.
   Reveal Solution Discussion Select an answer first
6. 10foundation · easy

   Which feature in Azure OpenAI Service allows you to define custom terms that the model should not generate?

   • ATemplates
   • BBlocklists
   • CSystem messages
   • DDeployments
   Reveal Solution Discussion Select an answer first