Stage 1 · 01 / 100
What AI Actually Is
The big idea
AI is a prediction machine — not a thinking machine.
Why you need to know this first
Everyone talks about AI like it's a digital brain. They say it “thinks,” “knows,” or “understands.” That framing will mess you up. It makes you trust answers you shouldn't, and second-guess outputs that are actually fine.
To use AI well, you have to know what it actually is. Not the marketing version — the real version. It only takes a minute, and every other page in this book builds on it.
What's actually happening
AI doesn't pull from a database. It doesn't look things up. It doesn't sit there forming a thought and then typing it out.
What it does is predict the next word. Given your question plus whatever words it has already written, it asks one question: “what word is most likely to come next?” It picks the best guess, adds it to the sentence, then asks again for the next word.
It does that thousands of times until the answer is done. That's the whole trick.
An example that lands
Ask ChatGPT: “What's the capital of France?”
It doesn't check a list. There's no lookup happening. What's actually going on is: it has seen the phrase “capital of France” followed by “Paris” millions of times in the stuff it was trained on. So when you ask, “Paris” has overwhelmingly the highest probability of being the next word.
It gives you “Paris.” And it's right. But it didn't know it. It predicted it.
Why this changes how you use it
Once you get this, everything else makes sense:
Remember this
AI doesn't think. It predicts. Remember that, and every weird thing AI does starts making sense.