Manage the Testing of AI-powered Business Solutions AB-100 Practice Questions (Page 5)

Questions 21–25

1. 21foundation · easy

   Which practice is a best practice for crafting effective prompts in Copilot?

   • AUsing vague language to allow creative freedom
   • BIncluding as much unrelated context as possible
   • CProviding clear and specific instructions
   • DAvoiding examples to prevent bias
   Reveal Solution Discussion Select an answer first
2. 22application · medium

   A QA team wants to use Copilot to help create test cases for an AI agent that handles refunds in Dynamics 365 Finance. They need a strategic approach to ensure thorough coverage. What should they do?

   • AAsk Copilot to generate all possible test cases automatically and use them without review.
   • BRely on Copilot to execute the tests and report results without defining expected outcomes.
   • CUse Copilot only to generate test data, not test cases, to avoid bias.
   • DProvide Copilot with a clear description of the refund process, edge cases, and expected outcomes, then review and refine the generated cases.
   Reveal Solution Discussion Select an answer first
4. 23expert · hard

   An AI agent in Dynamics 365 Customer Service handles returns and refunds. The team is evaluating its performance using metrics such as resolution rate, average handling time, and customer satisfaction. They notice that resolution rate is high, but customer satisfaction is low. What is the most likely cause?

   • AThe agent is resolving cases quickly but with low-quality or generic responses that frustrate customers.
   • BThe agent is being used by only a small number of customers, skewing the metrics.
   • CThe average handling time is too low, indicating the agent is not spending enough time on each case.
   • DThe customer satisfaction survey is being sent to the wrong customers.
   Reveal Solution Discussion Select an answer first
5. 24expert · hard

   A team has developed a custom AI model to predict customer churn in Dynamics 365 Customer Insights. The model shows high accuracy on the test set, but when deployed, it performs poorly on new data. The team suspects the model is overfitting. What validation approach should they have used to detect this?

   • AIncreasing the model's complexity to capture more patterns.
   • BCross-validation on multiple subsets of the training data.
   • CTesting the model on the training data to ensure it learns well.
   • DUsing a larger test set with more samples.
   Reveal Solution Discussion Select an answer first
6. 25application · medium

   A business analyst is crafting prompts for a Copilot that generates sales summaries in Dynamics 365 Sales. The team has noticed that the summaries are sometimes too verbose and miss key metrics. What prompt best practice should they apply?

   • AAdd multiple examples of different summary styles to confuse the model into being creative.
   • BAvoid any context about the audience to keep the prompt simple.
   • CUse a short, generic prompt like 'Summarize sales data' to let the model decide.
   • DInclude specific instructions about the desired length, format, and key metrics to include.
   Reveal Solution Discussion Select an answer first