Skip to content
Tech for Everyone Reader LogoTech for Everyone Reader

Chapter 2: How Robots Think and Learn

How Robots Think and Learn 💡

Section titled “How Robots Think and Learn 💡”

Step by Step!

In Chapter 1, we learned that robots see and hear with numbers. But how does a robot decide what to do? How does it learn new things?

Robots do not think the way you do. They do not feel or guess. They follow steps, and they learn from examples. Let’s find out how.

Maps for Thinking 🗺️

Section titled “Maps for Thinking 🗺️”

Robots “think” by following a list of steps. The list is called an algorithmalgorithm: A list of instructions that tells a computer exactly what to do, step by step.. An algorithm tells a computer exactly what to do, one step at a time.

You use algorithms every day! Here is an algorithm for brushing your teeth:

  1. Pick up your toothbrush.
  2. Put toothpaste on it.
  3. Brush the top teeth.
  4. Brush the bottom teeth.
  5. Rinse and spit.

If you skip a step or mix them up, it does not work. Robots are the same. They need every step, in order.

A Robot Playing a Game

Section titled “A Robot Playing a Game”

Have you played tic-tac-toe? When it is your turn, you look at the board and think. Should you block your friend? Should you try for three in a row? You are planning ahead.

When a robot plays tic-tac-toe, it imagines every move it could make. Think of a tree with branches. The bottom of the tree is the board right now. Each branch is a possible move. From each move, more branches grow. The robot checks the branches very, very fast and picks the one that looks best.

It may look like the robot is thinking like you. But it is really just following steps and checking choices, super fast.

Learning from Examples 🐟

Section titled “Learning from Examples 🐟”

Robots are not born knowing things. They must learn, just like you. One way a robot learns is by looking at lots of examples. The examples are called datadata: Examples and information used to teach a robot so it can learn..

Imagine you show a robot many pictures of fish. Some fish have a tag that says “Safe.” Some have a tag that says “Not safe.” The robot looks at all the pictures and searches for a patternpattern: Something that repeats in a certain way, like colors, shapes, or sounds..

  • Maybe all the red fish are not safe.
  • Maybe all the striped fish are safe.

After seeing many examples, the robot makes a rule. When it sees a brand-new fish, it uses the rule to predictpredict: Using what you have learned to guess what happens next. if the fish is safe. To predict means to make a smart guess.

The robot does not really understand what a fish is. It only knows the pattern of colors and shapes. That is how AI learns: lots of examples, then a rule.

You Can Learn Like a Robot

Section titled “You Can Learn Like a Robot”

Sort some blocks by color. Now you have made a rule: “red blocks go here, blue blocks go there.” If a friend looks at your piles, they can guess your rule. That is just what a robot does. It looks at many examples and finds the pattern.

Patterns Everywhere 🔁

Section titled “Patterns Everywhere 🔁”

AI is a pattern finder. Once you know that, you can spot AI all around you:

  • A tablet shows you a video you might like. It learned the pattern of what you watched before.
  • A phone fixes your spelling. It learned the pattern of how words are usually spelled.
  • A game gets harder when you get better. It learned the pattern of how you play.
  • A photo app finds all the pictures of your dog. It learned the pattern of your dog’s face!

Every time you use a device, you might be giving it a new example. That is why AI seems to “know” you. It does not know you like a friend does. It just knows your patterns.

When Robots Get It Wrong 🙃

Section titled “When Robots Get It Wrong 🙃”

Here is a secret: robots make mistakes. Sometimes lots of them! Here is why:

  • Not enough examples. If a robot only saw pictures of goldfish, it might not know a shark is a fish.
  • The wrong examples. If every “dog” picture had grass in it, the robot might think grass means dog!
  • A tricky new thing. A robot that learned in the daytime might get confused at night.

A robot cannot tell when it is wrong. It just follows its rule. That is why people check the robot’s work. If the answer looks silly, we fix the examples and teach it again.

So when a smart helper says something strange, do not worry. Just remember:

  • AI is a guesser, not a knower.
  • People check the guesses.
  • You can always ask a grown-up.

Robots think step by step. They learn from examples. They find patterns. And when they get it wrong, people help them learn better!

Chapter 2 Activity: Teach a Robot 🧠

Section titled “Chapter 2 Activity: Teach a Robot 🧠”

With a grown-up:

  1. Write an algorithm for making your favorite snack. Say every step in order. Did you forget one?
  2. Play robot. A grown-up follows your snack steps exactly. What went wrong? Fix the step!
  3. Sort some toys or blocks into two piles. Do not say the rule. Can your grown-up guess it?
  4. Find a pattern. Make a pattern with colors or claps. Can a friend guess what comes next?
  5. Draw a robot learning from three pictures. What is it learning?

Key Concepts Checklist

Section titled “Key Concepts Checklist”
  • I know an algorithm is a list of steps in order
  • I can say an algorithm for something I do every day
  • I know robots learn from examples called data
  • I know a pattern is something that repeats
  • I know AI finds patterns to make a smart guess
  • I know to predict means to make a smart guess
  • I know robots can make mistakes
  • I know people check a robot’s work and teach it again