AIThe Story of AI
Chapter Three · 1993–2011

Computers Start to Learn

A chess match. A robot vacuum. A car driving itself across the desert. The decade machines stopped following rules and started finding patterns.

For a long time, people built AI by writing giant rule books. They told the computer, "If THIS happens, do THAT." But there was a problem — real life has WAY too many "ifs" to write down! In the 1990s, scientists tried something new called machine learning. Instead of giving computers rules, they gave computers tons of examples and let the computers figure out the patterns themselves.

This worked AMAZINGLY well, especially because the internet was exploding with information. Suddenly there were millions of photos, web pages, and messages — the perfect "food" for hungry learning machines. New tools called Support Vector Machines and decision trees helped computers sort through all that data to make smart guesses, like "is this email spam or not?"

By the 2000s, AI started showing up in real life. It cleaned floors, drove cars across deserts, and even beat humans at games people thought only humans could win. This was the era when AI stopped being just a science experiment and started becoming part of everyday life.

Teaching by Examples (Not Rules)

Imagine you want a dog to know what a "ball" is.

That's machine learning — teaching by examples instead of rules!

§

The Decades AI Came to Life

1997
Deep Blue Beats Kasparov
IBM's Deep Blue beats world chess champion Garry Kasparov in a 6-game match. First time a computer beat a reigning world champion!
1998
Google is Born
Two Stanford students, Larry Page and Sergey Brin, create the PageRank algorithm, which becomes Google Search.
2002
Roomba Rolls In
iRobot releases the Roomba, the first home robot millions of families actually bought. Over 1 million sold in just a few years!
2004
DARPA Grand Challenge
The first self-driving car race is held in the Mojave Desert. NO car finishes — the best one only goes 7 miles!
2005
Stanley Wins!
Stanley, a self-driving car from Stanford led by Sebastian Thrun, drives 132 miles across the desert all by itself.
2006
Deep Learning Returns
Geoffrey Hinton publishes a famous paper bringing back "deep learning," setting the stage for today's AI.
2008
Voice Search
Google launches voice search on the iPhone, making speech recognition something regular people use every day.
2011
Watson Wins Jeopardy!
IBM's Watson beats champions Ken Jennings and Brad Rutter on TV — a bridge to the next AI era.
§

Three Stories From the ML Era

01

Kasparov Cried Foul

Kasparov got SO upset after losing to Deep Blue that he accused IBM of cheating with secret human help! (They didn't.) Deep Blue could look at 200 million chess positions every single second — no human brain can do that.

02

A Car With No Driver

Stanley the self-driving car had a regular Volkswagen Touareg body, but inside it had 5 laser scanners on the roof and 7 computers in the trunk. It drove 132 miles through the desert without any human touching the steering wheel!

03

The Vacuum's Cousin

The Roomba was actually built by the same company (iRobot) that made bomb-disposal robots for the military. So the little disc cleaning your living room is basically a peaceful cousin of a war robot!

1997 — IBM's Deep Blue defeats world champion Garry Kasparov. The first time a computer beat a sitting world chess champion.

Why Deep Blue Won

Chess positions evaluated per second — human brain vs. supercomputer.

~1 position Human (Kasparov) 200,000,000 Deep Blue (1997) POSITIONS / SECOND
Deep Blue could see 200 million possible chess moves every second. No human brain comes close.
Machine Learning is like training a real dog. Show it 100 balls and say "BALL!" each time — soon it figures out what a ball is, all on its own.
— The big idea that changed AI