Skip to main content
Machine Learning & Deep Learning

⏱ About 10 min10 XP

Machines Can Practice Too

You already know that kids learn by practicing. But here is something surprising: some machines can practice too! Not all machines — but a special kind. Today we are going to find out how a machine can practice and get better, just like you do.

Most Machines Never Change — But Some Do

Think about a toaster. Every single morning it works the same way. You put bread in, push down the lever, and toast pops up. The toaster does not get better at making toast over time. It does not learn from breakfast. Now imagine a different kind of machine — one that looks at thousands of photos of cats and dogs. The first time it tries to tell a cat from a dog, it might guess wrong a lot. But every time it guesses wrong, something inside it adjusts, just a tiny bit. After seeing millions of photos and fixing its mistakes over and over, it gets really good at telling cats from dogs. That machine practiced. That machine learned.

The Big Idea

Some machines can look at many examples, notice when they get things wrong, and adjust themselves to do better next time. That is how a machine practices!

Let us compare the two kinds of machines. Omar has a calculator. Every time he types 5 + 3, the calculator says 8. It will say 8 forever. It never gets faster or smarter. That is a regular machine — it follows fixed instructions and never changes. Omar also has a voice assistant on a tablet. At first it sometimes got his name wrong. But after he corrected it a few times, it learned exactly how to say his name. The voice assistant practiced from his corrections and got better. Two machines, two very different behaviors.

Match each machine to what makes it special.

Terms

Toaster
Cat-and-dog photo sorter
Calculator
Voice assistant

Definitions

Improves when you correct its mistakes
Gives the same answer every time, no learning
Gets better after seeing many examples
Always works the same way, never changes

Drag terms onto their definitions, or click a term then click a definition to match.

When a machine practices, it does not feel anything. It does not get tired or excited. But inside, tiny numbers are adjusting after every example — like tuning a radio a tiny bit each time until you find the right station. The more examples the machine sees, the better those tiny numbers get tuned. Scientists call this process training. A machine that trains on examples is learning!

A Word to Remember

Training is the word scientists use when a machine practices on lots of examples. A trained machine has seen so many examples that it can handle new ones it has never seen before.

What makes a learning machine different from a regular machine like a toaster?

What do scientists call the process when a machine practices on lots of examples?

Teach a Friend With Examples

  1. Choose a simple sorting game to play with a friend or family member.
  2. Collect 10 small objects from around your home — mix things like coins, buttons, paper clips, and erasers.
  3. Without telling your helper the rule, hand them objects one at a time and say 'yes' or 'no' based on a secret rule you made up (for example: round things are yes, square things are no).
  4. After each 'yes' or 'no,' your helper guesses the rule.
  5. Keep going until they figure it out!
  6. Talk about it: how is this like showing a machine examples so it can learn?