Chapter 7 · Part 4
Where this shows up
Telling humans from bots isn't a niche puzzle — it's load-bearing infrastructure for the whole internet. The ideas from this course run every time you sign in, check out or post, and they had a strange side effect: they helped build the AI they were meant to keep out.
Here's where the human-vs-bot fight actually happens.
Login & signup defense — block credential-stuffing and floods of fake accounts.
The same idea, many jobs
Every use is the same test in a different doorway:
- Logins, checkouts and posting are just the gate at different points — the goal is always a cheap-for-humans, costly-for-bots challenge.
- Modern defense leans on the invisible, behavioral approach — a risk score — because the puzzles themselves stopped working.
- And the twist from Chapter 3–4: solving them labeled the data that trained today's computer vision — the same CNNs and self-driving perception you can learn about elsewhere on the site.
The tradeoff to keep in mind
Behavioral detection works, but it means judging you by your mouse, timing and browsing history — real privacy cost. And it's an arms race with no final winner: bots adapt, defenders respond.
That's the course
You now know the whole story: the human-vs-bot problem, distorted text, the free-labeling trick, why traffic lights, the arms race, and the invisible CAPTCHA.
If you enjoyed this, the other courses cover the image and self-driving AI that your clicks helped train.