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.

• ✗

  Supervised learning requires human oversight; unsupervised runs automatically

  Why it's wrong here

  This is incorrect because the distinction between supervised and unsupervised learning is not about human involvement at runtime. Both paradigms can run fully automatically once the training process is configured. The defining difference is that supervised learning requires training data that includes input-output label pairs, whereas unsupervised learning operates on unlabeled data to discover hidden structure on its own.

• ✓

  Supervised uses labeled training data; unsupervised finds patterns in unlabeled data

  Why this is correct

  This is the correct definition. Supervised learning builds a model by learning a mapping from input features to known output labels, as in image classification or house-price regression. Unsupervised learning instead finds patterns, groupings, or latent structure in data that has no preassigned labels, such as customer segmentation through clustering. The presence or absence of labels in the training dataset is the core differentiator between the two paradigms.

• ✗

  Supervised is for images; unsupervised is for text

  Why it's wrong here

  This is wrong because neither learning paradigm is restricted by data modality. Supervised learning can be applied to text (e.g., spam detection) and images (e.g., object recognition), and unsupervised learning similarly works across text (e.g., topic modeling) and images (e.g., image clustering). The choice of paradigm depends on whether labeled targets are available and what question you are trying to answer, not on the type of data you have.

• ✗

  Supervised is older and less accurate than unsupervised

  Why it's wrong here

  This is incorrect because both paradigms have been researched for decades and their relative accuracy depends on the specific task and data, not on the paradigm itself. When high-quality labeled data exists, supervised learning typically achieves higher predictive accuracy by using ground-truth labels, while unsupervised learning excels at uncovering hidden structure when labels are missing or too costly to obtain. Thus, describing one as universally 'older' or 'less accurate' than the other is a false generalization.

Go deeper

Related to this question