3.1 Describe How to Proactively Overcome Foundation Model Limitations. GENERATIVE-AI-LEADER Practice Questions (Page 2)

Questions 6–10

1. 6foundation · easy

   A team wants to ensure that their generative AI application benefits from improvements made by the model provider, such as bug fixes and performance enhancements, without manual intervention. Which practice is most relevant?

   • AVersioning
   • BAutomatic model upgrades
   • CDrift monitoring
   • DPerformance tracking
   Reveal Solution Discussion Select an answer first
2. 7application · medium

   A customer service team is using a generative AI chatbot to handle refund requests. The chatbot sometimes provides refund policies that are no longer valid. The company updates its refund policy quarterly. The team needs the chatbot to always reflect the current policy. What is the most efficient approach?

   • AManually update the prompt with the new policy text every quarter.
   • BUse RAG to ground the chatbot in the latest policy document stored in a knowledge base.
   • CFine-tune the chatbot on the new policy every quarter.
   • DUse a model with a more recent training cutoff date.
   Reveal Solution Discussion Select an answer first
4. 8foundation · easy

   A company wants to build a customer support chatbot that can answer questions about its own product documentation, which is updated frequently. Which Google-recommended practice is most suitable to ensure the chatbot provides accurate, up-to-date answers?

   • AFine-tuning the model on the documentation
   • BHuman-in-the-loop for every response
   • CPrompt engineering with generic instructions
   • DRetrieval-augmented generation (RAG)
   Reveal Solution Discussion Select an answer first
5. 9expert · hard

   A company has a generative AI model that summarizes financial reports. The model is critical for compliance, and the team must ensure it remains accurate over time. The team has limited resources and cannot afford to retrain frequently. Which strategy best balances accuracy and resource constraints?

   • AImplement continuous monitoring with key performance indicators and use RAG to ground the model in current financial data.
   • BFreeze the model version and rely on its initial training data.
   • CUse a larger model that can handle all financial data without updates.
   • DFine-tune the model monthly on the latest financial reports.
   Reveal Solution Discussion Select an answer first
6. 10application · medium

   A social media company uses a generative AI model to moderate comments. The model sometimes flags neutral comments as toxic, especially for certain dialects. The team wants to reduce this bias. What is the most appropriate mitigation?

   • ADisable the moderation model and use human moderators exclusively.
   • BIncrease the model's sensitivity threshold to reduce false positives.
   • CEvaluate the model's performance across different dialects and use the findings to improve the prompt or retrain with balanced data.
   • DUse a larger model that is less likely to be biased.
   Reveal Solution Discussion Select an answer first