AI, Machine Learning and Deep Learning

Audience

ages 10–15

Duration

30 to 40 min

Adventure

1/8

Badge to earn

Systems Cartographer

Today's mission

Build a map that tells apart a programmed rule, a learned model and a deep network.

Proof of success: explain "AI, machine learning and deep learning", complete "The Ability Circles" and name the limit of the analogy.

The story

At the science club, Inès wants to call "AI" every object that reacts. Malik lays down three cards: a timer follows a written rule, a seed sorter adjusts its parameters from labeled examples, and an image system uses many layers. All three produce a result, but they don't get there the same way. Their investigation is about asking where the rule comes from and how the result was checked.

What is technically true

Watch the mechanism, not just the result. The three markers below let you check each step. Here, your anchor is "AI, machine learning and deep learning".

• Artificial intelligence is a large field; some AI systems rely on rules and don't learn at all.
• Machine learning adjusts the parameters of a model from data and a measured objective.
• Deep learning is a part of machine learning that uses networks with several transformations; it guarantees neither truth nor understanding.

The vertical map

The map begins with the problem, distinguishes where the rule comes from, then ends with a measurement and a human decision. Read the diagram from top to bottom: each arrow announces a transformation or a check, never a thought inside the machine.

A useful analogy — and where it breaks

The analogy: Picture three nested boxes: AI on the outside, machine learning inside, deep learning even further at the center.

Where it breaks: The fields overlap in real-world uses and a box says nothing about the quality, cost or safety of a system. An analogy helps you get started; it never replaces the data, the calculations or a test.

Screen-free activity — The Ability Circles

Materials: three loops of string, six cards showing fictional systems, and a pencil
Suggested time: 10 to 20 minutes.

1. Place the big AI loop, then the machine learning and deep learning loops inside it.
2. Sort a timer, a rules-based tree, a learned classifier, a CNN, a spreadsheet and a fictional chatbot.
3. For each card, write the evidence that justifies its place instead of trusting the name.
4. Move any card whose evidence is too weak to a "to double-check" zone.

What you should notice: a system can automate a task without learning, and the word AI is not enough to know its mechanism.

Optional challenge

Add a seventh card: a thermostat that tunes its own parameters. Explain what information is missing before sorting it.

Safety, fairness and human choice

An "AI-powered" label can be pure marketing. Before using a tool, look for what data it receives, what error was measured, and who can push back on its output.

Responsible question: In what situation would a simple rule be clearer, cheaper and safer than a learned model?

Mini-quiz with answers

1. Does every automatic system learn?

   Answer: No. A timer or a chain of conditions can be entirely programmed by hand.

2. Where does deep learning sit?

   Answer: Inside machine learning, which itself sits inside the wider field of AI.

3. Does a deep network understand an image like a person does?

   Answer: No. It transforms numbers and learns measured associations; the word "understand" would be misleading here.

Competency badge

Systems Cartographer — You earn this badge if you can explain the diagram without reading, show the result of the activity and kindly correct a wrong answer.