What Is Machine Learning

What Is Machine Learning

Step 1 of 12

Ordinary code states the rule, and the machine applies it to whatever comes in.

The Idea

Ordinary code states the rule and the machine applies it. Machine learning flips that: you supply examples paired with correct answers, and training fits a rule that reproduces them. That rule lives in numbers — parameters — not in lines anyone can read.

Real-World Example

A postal sorting centre reads handwritten postcodes. Nobody can write down what a hand-drawn seven looks like across every hand on earth. Feed the sorter millions of scanned, human-labelled envelopes and it learns the shapes anyway.

The Tradeoff

You give up readability and control. A fitted rule works only where new inputs resemble the training data, degrades quietly when the world shifts, and costs labelled data plus compute up front. When the rule is short and knowable, plain code stays cheaper, exact, and auditable.

Your turn

Put the steps in the right order.

1. 1Deploy the model and watch how it behaves on live inputs↑↓
2. 2Collect examples and label each one with the correct answer↑↓
3. 3Train: adjust the parameters until predictions match the labels↑↓
4. 4Hold back a slice of the data that training never sees↑↓

Check order

Mini quiz

1 / 3

In machine learning, where does the rule come from?

• AA programmer writes it out explicitly
• BIt is fitted from labelled examples
• CThe compiler infers it from types
• DIt is picked at random each run