Lesson 2 · AI Explorer
From patterns to guesses.
When an AI says "that's a cat", it isn't reporting a fact. Underneath, it calculated something closer to:
87% cat, 9% dog, 4% something else.
It then showed you the top one. The uncertainty was always there — it just got hidden.
That's why AI can be wrong without any error message. It wasn't ever certain. It gave you its best guess, formatted like a statement of fact.
The clearest example is on your phone right now.
Type "I am going to the" and it suggests market, school, shop. It has no idea where you're going. It has seen millions of sentences and knows which words usually follow.
Large AI chat systems do a more sophisticated version of exactly this: predict what text most plausibly comes next.
That explains a lot. It's why they write fluently — fluency is the thing being predicted. And it's why they can state something false so smoothly: a plausible-sounding sentence and a true one look identical to a system predicting plausibility.
5 more parts in this lesson
Walkthroughs, code labs, mini-games and the checkpoint quiz unlock when you buy the course.