AI-300 Practice Question 68: Intent Resolution Evaluator

Microsoft Machine Learning Operations Engineer Associate (AI-300) · Test 7 · question 8 of 10

Implement generative AI quality assurance and observabilitymedium

A customer support agent sometimes misunderstands what the user is asking for, and then calls tools for the wrong task. The team wants a built-in agent evaluator that scores whether the agent correctly identified the user's intent. Which evaluator should they use?

Select one answer
Show answer and explanation

Answer:

  • C. The Intent Resolution evaluator

Foundry's agent evaluators include Intent Resolution, which measures whether the agent correctly identifies the user's intent. Process evaluators such as Tool Call Accuracy check the calls that follow, so Intent Resolution is the one that targets the misunderstanding itself (option C).

- A is wrong because groundedness checks whether a response is supported by the provided context. It does not judge how the agent interpreted the request. - B is wrong because fluency scores the readability and grammar of the text, and an answer can be fluent while addressing the wrong intent. - D is wrong because tool call accuracy judges whether the calls made were correct and used the right parameters. It does not score the interpretation of intent itself.

Reference: Agent evaluators

Practice Test 7 ›

Study guides for AI-300