Skip to content
BENIBIX Learn

Prediction (AI)

AI🧭 FoundationsAI Basics

Also known as Model Output, AI Prediction

The answer a model gives for one input. It is its best guess, not a fact.

Think of it like

Think of it like a guess about rain. You look at the sky and say what you think.

Example

Lucy's email app predicts that a message is spam, with a score of 0.97. It moves the message to the spam folder.

Why it matters

Every prediction is a guess with some chance of being wrong. Good apps plan for the wrong ones.

Definition

A prediction is the answer an AI model gives for one input. It is called a prediction even when nothing is about the future. A model that labels a photo as a cat is predicting the label. A chatbot predicts the next word to write. Most predictions come with a score that says how sure the model is.
The word cat at the end of the bottom row is the prediction. It is the model's best guess for that photo.

Common mistake

Treating a model's answer as a known fact. A high score means the model is sure, not that it is right.

Key takeaways

  1. 1
    Best guess. Not a fact, a likely answer.
  2. 2
    Any output. A label, a number or a word.
  3. 3
    Comes with a score. How sure the model is.

Learn it properly