Part of the Pristone Academy AI Technical Track

Intro to artificial intelligence — teaching machines to think.

What AI actually is, where it came from, and why it suddenly became impossible to ignore.

200,000 FT

The Big PictureWhy it matters

AI is the project of getting machines to do things that normally require human intelligence.

For most of history, computers were calculators with good PR. You gave them precise instructions — add these numbers, sort this list, open this file — and they followed them perfectly, blindly, endlessly. The machine was fast; the intelligence was entirely yours.

Artificial intelligence is the attempt to change that bargain. Instead of writing every rule by hand, AI researchers asked: what if we could give a machine a goal and let it figure out the rules itself? What if it could learn to recognize a face, translate a sentence, or win a board game — the way a person does, through experience?

Think of the difference between teaching a child to recognize a cat by writing out a rule book (“four legs, pointy ears, whiskers…”) versus showing them a thousand pictures of cats until the pattern clicks. AI is the second approach, applied to machines.

The dream is old — the term “artificial intelligence” was coined at a 1956 Dartmouth conference — but the payoff is recent. Modern AI reads your X-rays, understands your voice commands, suggests your next line of code, and recommends the next video you can't stop watching. It is the most economically significant technology since the internet, and it is still accelerating.

What AI is not

AI is not magic, and it is not a single thing. It is a family of techniques — search, logic, statistics, and lately neural networks — united by the goal of automating tasks that require judgment. The techniques are mundane math. The results are sometimes astonishing.

30,000 FT

The LandscapeHow it's organized

AI is a country. Machine learning is a city in it. Deep learning is a neighborhood in that city.

The single most useful map of the field is a set of nested regions. Artificial intelligence is the broadest — any technique that makes machines act intelligently, including old-school approaches like hand-written logic and search.

Inside it sits machine learning: systems that learn patterns from data instead of being explicitly programmed. This is where nearly all modern progress lives. And inside that sits deep learning — machine learning built on neural networks with many layers, the engine behind image recognition, speech, and the large language models you've been hearing about. Every news headline about AI is really a headline about that innermost neighborhood.

The three ways a machine learns

Almost all machine learning falls into three styles. Supervised learning trains on labeled examples — show it thousands of emails marked “spam” or “not” and it learns to sort new ones. Unsupervised learning gets no labels and finds structure on its own — grouping similar customers, say. Reinforcement learning learns by trial and error against a reward, the way a program gets better at a game by playing millions of rounds.

So, is it AI? Test your instinct.

The line between “AI” and “just a program” comes down to one question: did a human write the rules, or did the machine learn them? Try a few cases below and check your reasoning against each answer.

Spam filter

Reads your email and moves suspicious messages to Junk.

Calculator

Multiplies two numbers together exactly as instructed.

Face unlock

Your phone recognizes your face in different lighting, with glasses, or at an angle.

Thermostat

Turns the heat on when the temperature drops below 68 °F.

Recommendation engine

Suggests your next show based on what you and similar viewers watched.

Route planner (core)

Finds the shortest path between two addresses using a road graph.

A concrete example: spam detection

A spam filter is the cleanest case. Nobody could write rules for every trick a spammer invents. Instead, the filter is shown millions of labeled emails and learns the patterns of spam itself — then keeps adapting as the spam evolves. Rules learned from data, not hand-coded: that's the whole idea of AI in one familiar tool.

0 FT

What to Carry OffThe summary

Three things to keep about AI.

Learned, not hand-coded

What makes something AI is that its behavior is learned from data or experience, instead of every rule being written out by a person.

A family, not a thing

AI is many techniques — search, logic, statistics, neural networks — sharing one goal: automating tasks that need judgment.

Three ways to learn

Supervised (labeled examples), unsupervised (find structure), and reinforcement (trial and error against a reward) cover most of it.

Old idea, recent payoff

The concept dates to 1956; data, GPUs, and better algorithms are what finally made it work at scale.

The one sentence to remember

Artificial intelligence is machines learning to do judgment-heavy tasks from data and experience, rather than following rules a human wrote out in advance.

This is the on-ramp. If you want a guided path from “I understand the map” to actually building things, that's what a 1:1 session is for.

Frequently asked questions

Artificial intelligence is the project of getting machines to do things that normally require human intelligence — recognizing a face, understanding speech, translating a sentence, winning a game. Instead of a human writing every rule by hand, the machine learns the rules itself from examples.

Think of nested regions. AI is the whole country: any technique for automating tasks that need judgment. Machine learning is a large city in it: systems that learn patterns from data rather than being explicitly programmed. Deep learning is a neighborhood in that city: machine learning built on many-layered neural networks.

Mostly in three ways. Supervised learning trains on labeled examples (emails marked spam or not). Unsupervised learning finds structure in unlabeled data (grouping similar customers). Reinforcement learning learns by trial and error against a reward (a program getting better at a game by playing it).

No. A calculator or a thermostat follows fixed rules a human wrote — fast and reliable, but no learning or judgment. It only counts as AI when the system learns its behavior from data or experience instead of having every rule hand-coded.

The ideas are old — the term was coined in 1956 — but three things recently lined up: enormous amounts of data, cheap and massively parallel computing power (GPUs), and better learning algorithms. Together they let neural networks reach a scale that turned decades of slow progress into a sudden leap.

K-Means Clustering

How a machine sorts a pile of points into clean groups with no labels.

Principal Component Analysis (PCA)

How a machine keeps the few directions that matter and throws away the rest.

Attention Is All You Need

The self-attention idea behind every modern transformer and LLM.

The AI Layer Shift Framework

How AI redistributes work across the layers of a job — and where humans stay scarce.

Curious who's behind these lessons? See the proof of work.