Course 0 · Lesson 04/10 · Interactive guide

AI, ML, Deep Learning, and Generative AI

These names are related, but they are not interchangeable. Open each branch to see what it takes in, what it produces, and where it fits.

Lesson
04 / 10
Time
About 9 minutes
Level
Beginner · 0/5
Prerequisite
Data Is AI's Food

Start with the widest circle

AI names a whole field, not one method

Picture a large toolbox labelled AI. It contains different tools. Some follow rules written by people, some learn patterns from examples, and some generate new text, images, or sound from patterns learned during training.

The widest circle

Artificial Intelligence (AI)

A field and a collection of systems that perform tasks such as classifying, planning, predicting, or choosing an action.

Two important branches

Rules and learning from data

AI can use symbolic rules supplied by people or machine learning fitted from examples. Many real systems combine both.

Try the explorer

Open the AI family map one branch at a time

Choose a box with a pointer, or press Tab and then Enter/Space. The definition panel changes with your selection.

The wide field containing every branch

Artificial Intelligence (AI)

What it is
A field containing several ways to build systems that classify, predict, plan, or choose actions.
Input → output
Data, state, or rules → a decision, answer, or action
Examples
Route planners, spam filters, and game-playing programs

Spot it in real life

One product can combine several branches

A

An email filter

An ML model might score how closely a message resembles spam. Human-written rules can then decide which folder receives it.

B

A route planner

A rule-based search algorithm can find possible routes, while ML may estimate travel time from earlier traffic patterns.

C

A text-to-image tool

A deep-learning model learns patterns linking text and images, then produces a new image from a prompt. That makes the task generative.

Use the terms precisely

Real boundaries are not always tidy circles

TermCore methodTrade-off
Symbolic AIRepresents knowledge with rules, logic, or symbolsRules can become difficult to write and maintain in a complex world
Machine LearningFits model parameters to example dataResults depend on the quality and coverage of that data
Deep LearningUses many-layer neural networks to learn representationsOften needs more data and computation than a simpler model
Generative AILearns a data distribution and produces new samplesA convincing output can still be wrong or unsuitable for the context

Many modern generative systems use deep learning, but “generative” describes the task, while deep learning describes a modelling method. The two labels should not be treated as synonyms in every case.

Remember three things

Keep this map before moving on

1

AI is the widest field. It contains systems based on rules as well as systems fitted from data.

2

ML sits inside AI, and deep learning sits inside ML. A deeper method is not necessary for every task.

3

Generative AI focuses on producing new outputs. Those outputs still need review that matches their risk.

Three-question check

Read the map without scrolling back

1. Which statement best describes the relationship?

Choose an answer to see the explanation.

2. Can a system that follows “if…then…” rules without training still count as AI?

Choose an answer to see the explanation.

3. Why should a generative AI answer still be checked?

Choose an answer to see the explanation.