The type of AI that focuses on classifying or identifying content based on preexisting data is generally called supervised learning AI, often implemented through discriminative models.

Core idea in simple terms

  • Supervised learning uses labeled historical data (examples with known answers) to learn patterns.
  • Once trained, it can take new, unseen inputs (emails, images, documents) and classify them into categories like “spam / not spam” or “cat / dog.”

Why supervised learning fits this question

  • It explicitly relies on preexisting labeled data: each training example is tagged with the correct class before learning starts.
  • The model’s goal is to map inputs to labels as accurately as possible, making it ideal for recognition and identification tasks such as sentiment analysis, fraud detection, and medical diagnosis support.

Role of discriminative models

  • Many supervised models are discriminative : they learn the boundary between classes (for example, what separates “spam” from “not spam”).
  • In multiple-choice style explanations online, this “AI that focuses on classifying or identifying content that is based on preexisting data” is often labeled as discriminative AI, in contrast to generative AI that creates new content.

Quick comparison (context)

  • Generative AI: learns to model how data is distributed and can generate new examples (like chatbots or image generators).
  • Supervised/discriminative AI: learns decision boundaries and focuses on labeling or identifying existing data points based on what it has seen before.

TL;DR: In exam-style wording, the expected answer is usually “discriminative AI,” but more broadly in modern AI terminology, this behavior is achieved through supervised learning models trained on labeled, preexisting data.