AI-900 Practice Question: Describe Artificial Intelligence workloads and considerations

Answer analysis

Option-by-option breakdown

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

• ✗

  Replacing all human data labelers with ML models

  Why it's wrong here

  Replacing all human data labelers with ML models removes the human role entirely, making it a fully automated process rather than human-in-the-loop. While ML models can label data at scale, they are prone to systematic errors that humans would ordinarily catch, especially for ambiguous or out-of-distribution examples. The 'loop' in human-in-the-loop specifically means that a human remains in the decision path, reviewing and correcting model output. Without human review, the system loses a critical quality assurance safeguard and becomes less trustworthy for training future models.

• ✓

  Using ML to pre-label data while routing uncertain cases to human reviewers for quality assurance

  Why this is correct

  In a human-in-the-loop labeling workflow, an ML model first pre-labels a dataset, then the system routes low-confidence predictions to human reviewers. This blends automation with expert judgment: the ML model handles straightforward cases quickly, while humans focus on ambiguous or uncertain instances where the model is less reliable. The human corrections are often fed back into the model for retraining, which improves future pre-labeling accuracy and maintains overall data quality.

• ✗

  Requiring all data to be labeled by humans without any ML assistance

  Why it's wrong here

  Requiring all data to be labeled by humans without any ML assistance is the antithesis of a human-in-the-loop design. While it preserves human judgment, it forfeits the speed and cost benefits that ML pre-labeling provides. In Azure ML, human-in-the-loop is explicitly defined as the collaboration between people and an ML model—not a purely manual process. Without ML assistance, labeling becomes a slow, expensive bottleneck, especially for large datasets, and humans are still susceptible to fatigue-induced errors.

• ✗

  Using a loop in Python code to automate the labeling process

  Why it's wrong here

  Using a Python loop to automate labeling is a software engineering approach, not a human-in-the-loop process. A for loop that programmatically assigns labels lacks both the adaptive intelligence of an ML model and the oversight of a human reviewer. Human-in-the-loop requires semantic feedback—humans correcting uncertain cases—whereas a Python loop merely executes a fixed, deterministic set of rules. It may be part of an automation script, but it does not embody the human-supervised machine learning lifecycle that defines human-in-the-loop labeling.

Go deeper

Related to this question