Unit 8 · AI & the Impact of Computing · Lesson 8.1

What Machine Learning Actually Is

Every program you’ve written this year followed rules you wrote and could explain. Machine learning flips that: you give it examples, and it works out the rules itself — rules nobody wrote down.

Big Idea 5 — Impact of ComputingCED 5.1 · 5.3 · Unit 8 opener

This unit goes beyond the AP exam on purpose, because you live with AI every day. Every page marks what’s AP-TESTABLE and what’s CONTEXT — NOT ON THE AP EXAM. Trust the labels: exam-tagged questions only test the green material.

Before this lesson

No reading — this page is the reading. Bring one of your Unit 3 programs to mind: every if in it is a rule you wrote.

Where We Are

People→ Bits→ Programs→ Internet→ Security→ Big Data→ Impact

Impact lights up — the last region. It loops back to People, where the map began on Day 1.

The Hook

Teach a computer what a cat is — without telling it. Your teacher trains an image classifier live: a few dozen webcam photos of two objects, no rules, no code. Five minutes later it tells them apart. Nobody typed a single if.

The Idea

	Traditional programming	Machine learning
You give it	rules (your code)	examples (labeled data)
It gives back	answers	rules it worked out itself (a model)
Can someone explain the rules?	yes — the programmer wrote them	often not — they came from data

CONTEXT — NOT ON THE AP EXAM

Training data is the set of labeled examples. Training produces a model — the learned rules. Using the model on something new gives a prediction.

AP-TESTABLE

What the exam does say: machine learning and data mining have enabled innovation in medicine, business, and science — and information discovered this way has also been used to discriminate against groups of people. Both halves, every time.

Try It — Watch a Rule Get Learned

CONTEXT — NOT ON THE AP EXAM

A tiny learner sorts animals into cat or dog using one feature: weight. It averages each group’s weights and puts its rule halfway between. Add examples and watch the rule move — you never write it.

add a training example:

NEW ANIMAL (kg)

Add the chihuahua and watch the dog average drop and the dividing line move with it. Now add the big cat. Nobody edited the rule — the model learned exactly what its examples taught it, and that’s tomorrow’s whole lesson.

Vocabulary

artificial intelligence (AI) Impact: Computer systems that do tasks we associate with human intelligence, like recognizing images or language.
machine learning Impact: A kind of AI where the computer finds patterns in example data instead of following rules a person wrote.
training data Impact: The labeled examples a machine-learning system learns from. CONTEXT — NOT ON THE AP EXAM
model Impact: The rules a machine-learning system produces from its training data. CONTEXT — NOT ON THE AP EXAM
prediction Impact: The model’s output for a new input it hasn’t seen. CONTEXT — NOT ON THE AP EXAM

Check

Five questions. Look at each tag: BI5 items are exam-style; context items aren’t on the exam.

0 of 5 answered

Impact Check

When nobody can say why

If a teacher’s rule decided your grade, you could ask for the rule. If a model decided it, the honest answer might be “the training data led it there.” Many people argue that decisions about people — jobs, loans, school placement — should come with an explanation a person can check. Should they?

Connections

Came from 2.12 — every if you wrote was a rule you could explain. Machine learning’s rules come from data instead.

Came from 5.1 — data is the raw material. Here, it becomes the program.

Next in 8.2 — the model can only learn what the examples show. What happens when the examples are lopsided?

Returns in 8.3 — you build a learner like this one yourself.