AI-900 Practice Question: Describe features of Natural Language Processing workloads on Azure

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  A tab showing the evaluation scores given by users to the AI's responses in production

  Why it's wrong here

  User feedback scores are a form of product analytics, not model evaluation. In Language Studio, the evaluation tab reports how a custom model performs on a held-out test set with known labels, while user ratings in production reflect satisfaction and usability, not ground-truth accuracy. Confusing these conflates subjective experience with objective, label-based performance measurement.

• ✓

  Performance metrics (precision, recall, F1, confusion matrix) on held-out test data for custom models

  Why this is correct

  Language Studio's evaluation for custom models reports precision, recall, F1 score, and a confusion matrix computed on held-out test data that was not used during training. These metrics break down per class or entity type, letting you identify specific weak spots—such as an entity category with low recall—so you can collect more targeted training data and retrain the model.

• ✗

  Environmental evaluation showing the compute carbon footprint of model training

  Why it's wrong here

  An environmental evaluation of compute carbon footprint belongs to sustainability reporting, not to machine learning model evaluation. Carbon emissions from model training are tracked with Azure's emissions dashboard or sustainability tools, whereas Language Studio's evaluation tab reports statistical accuracy measures like precision and F1. These are different methods and serve entirely different purposes, so this option describes the wrong kind of "evaluation."

• ✗

  A compliance evaluation checklist verifying the model meets data privacy requirements

  Why it's wrong here

  A compliance evaluation checklist for data privacy is a governance and security tool, not a predictive performance evaluation. Language Studio's evaluation tab measures the correctness of a custom model's predictions against labeled test data, using metrics such as precision and recall. Compliance checks, by contrast, verify adherence to regulatory requirements and are typically handled through Azure Policy or Compliance Manager, not the model's evaluation report.

Go deeper

Related to this question