The first rung of Module 6 · meaning gets a home address
[0.8, −0.3, …]
Every word you type gets snatched, gutted, and turned into an arrow of numbers before an AI does one lick of “thinking.” Words used the same way end up pointing the same direction — so meaning becomes something you can actually measure. Wildest part: you can do arithmetic on it. king − man + woman lands you right on top of queen. No, really — you're about to watch it happen.
Why this quietly runs your life
Every time you type a sentence to an AI — ChatGPT, Claude, the autocomplete on your phone, a search bar — this is the very FIRST thing that happens, before any “reasoning,” before any answer, before anything you'd call thinking. Your words get chopped up and swapped for arrows of numbers. Everything downstream — including the next rung, attention — is just more arithmetic on those same arrows.
That's not a metaphor. It's not “kind of like” math. It is math — and you're about to watch it happen below.
Step 1: computers can't read a single word
Show a computer the word king and it sees… nothing. No crown, no throne, not even a vibe. A computer only ever does one thing: math on numbers. So the very first move of every AI sentence is non-negotiable — every word needs a number address, or nothing downstream can even begin.
Step 2: good news — you already met the address format
Back in Module 2 you met the vector: a list of numbers that's secretly an arrow, pointing somewhere with some length. An embedding is just that same arrow, aimed at a word. king becomes something like [0.8, −0.3, 0.5, …] — in a real AI, hundreds of numbers deep. Same arrow you already own. New job.
Step 3: place the arrows so meaning becomes a location
Nobody sits there hand-assigning these numbers. An AI reads mountains of real sentences and notices which words keep showing up in the same kind of slot — “the ___ barked,” “the ___ meowed.” Words that share a neighborhood get placed near each other. Distance on the map is similarity of meaning.
Click a word below and watch its actual nearest neighbors light up.
Click a word to see who it’s closest to.
Step 4: one rung up, the arrows start talking to each other
Right now every word's arrow just sits there alone, ignorant of its neighbors in the SENTENCE (not the map). The word “it” has no idea what it refers to yet. Next rung, arrows learn to glance at each other and borrow meaning — that's attention, and you already own both ingredients it needs: this map, and the dot-product similarity meter from Module 2.
Prove it · get one wrong and I’ll walk you through it
Two quick ones and the rung is yours.
On the meaning map, kitten sits closest to…
king − man + woman lands almost exactly on queen. That proves…
Answer both to unlock the next rung ↓
Words, mastered as directions.
You didn't just learn a fun fact about a king and a queen — you learned that meaning can be a location, and locations support arithmetic. Anywhere this shows up again, you skip it.
Next rung → Attention