a visual course

Why Captchas Show
You Traffic Lights

(CAPTCHAs)

The web's human-vs-bot test — and why solving it quietly trains AI — in about half an hour.

A free, visual course on how CAPTCHAs work: the human-vs-bot test, the reCAPTCHA trick that turned solving puzzles into free data-labeling, why you click traffic lights, and how detection went invisible once AI learned to see.

in the real world

The same human-vs-bot ideas guard logins and checkouts everywhere, fight fraud and spam — and, surprisingly, helped label the data that trained today's AI.

Start at the gate →

Part 1 — Telling humans from bots

1

Are you a robot?

Spam, fake accounts and scalpers are mostly bots — so the web needs a quick test only a human can pass.

2

The first test: read this squiggle

Warped, noisy text was easy for people and hard for the machines of the day — the original CAPTCHA.

Part 2 — The clever trick

3

Free labor: you are the labeler

reCAPTCHA slipped an unknown word in next to a known one — so passing the test also digitized books, for free.

4

Why traffic lights

The same trick, now with images — and the labels you produce are training data for maps and self-driving cars.

Part 3 — The arms race

5

The arms race

The better computer vision got, the easier the puzzles became for bots — until they barely told anyone apart.

6

The invisible CAPTCHA

So detection moved to behavior: how you move the mouse and browse becomes a quiet risk score — the puzzle is just the fallback.

Part 4 — Out in the world

7

Where this shows up

Guarding logins and checkouts, fighting fraud and spam — and labeling AI's training data.