What happens when you press send

What you'll walk away with: the machine in one sentence you can check, the x-ray of a single send, and the reason a brilliant draft and a confidently invented citation are the same behavior.

You need: both tabs from the last lesson, about 20 minutes, and paper for one sentence. The two demos run in your browser.

Write your prediction

Before anything else, write one sentence, on paper: what do you think happens in the moment you press send? If you have never used one of these tools, write what you imagine. Keep it; you will check it against the machine before the end.

And here is the machine in one move, because you have met a tiny version of it already. The suggestions that hover above your phone's keyboard are a small pattern machine with one job: read the last few words you typed, propose the next little chunk. This is that, given whole documents instead of a few words — a scale that changes what the move can do. It reads in a bunch — a whole email thread, a whole contract, an afternoon of notes — and it writes out a bunch, making the same move again and again, each chunk chosen to follow everything before it, until a whole answer stands on the page. "Super-charged autocomplete" is the right first idea, not a joke at the machine's expense. Hold onto it; this lesson makes it precise — including the parts it does not prepare you for.

Where it came from

The model was not programmed rule by rule; it was trained, and everything else follows from that word.

Act one. A program was given a colossal share of human writing — books, contracts, articles, code — and one job: read a passage, guess the next small chunk, check the guess, adjust, repeat. Trillions of times. Past spelling and grammar, the only way to keep improving is to absorb the deep structure of what it reads — how a contract clause is built, how an argument concedes and turns. Competence is what next-chunk prediction extracts from competent text. That is why predicting words can do real work: it practiced on the work product of people who could.

When training ends, everything it learned freezes — and what it learned is odds: for any run of text, which chunk tends to come next. Those frozen odds are the instincts you will watch in a minute, and nothing you ever type changes them; pasting a memo into a chat does not teach the model your client's affairs. Whether your text is kept to help train future models is a separate thing — an app setting, on by default in some products, and lesson 12 shows you where it lives.

Act two. A machine trained only to continue the internet's writing will continue anything, so the companies shaped it: people were hired to write thousands of ideal answers — helpful, structured, confident — and the model was trained again to continue in their voice. What answers you is a statistical simulation of those hired expert writers, which is why it reads like a capable colleague and never like a page of search results. One consequence of that shaping runs through everything on this site; you are about to watch it happen.

Meet the pattern machine

Now, the sentence you wrote at the start. If it described looking something up, you are in good company — it is the most expensive wrong idea in this subject, and wrong in a specific way: the thing answers like a search engine, so we assume it works like one. It doesn't. It predicts the next chunk of text, then the next, then the next, and no plan survives beyond what gets written. When a product shows you its "thinking," that is more text from the same move; there is no small researcher behind the glass deciding what you deserve to know.

Here is the whole loop, drawn:

THE LOOP — WHAT EVERY SEND ACTUALLY SENDSSCHEMATIC — THE SHAPE, NOT THE SIZES

THE WINDOW

• the app's instructions
• the app's memory file
• the chat so far
• your files
• your newest message

MODELfrozen odds — nothing you type changes them

NEXT CHUNKone small piece of the reply

⟲ each chunk is appended and the odds run again until the reply ends. On your next send, the pile rides again — as much of it as fits, the app trimming or condensing the rest — finished reply included; no model runs between sends.

Don't take the origin story on faith. What follows is an x-ray of one send, and a simulation: hand-authored numbers, no model behind the glass. The toy's own face says SIMULATED, and its reference entry, the pattern machine, says so in its first line. The shape is the real thing; watch the shape.

Press SEND and follow it through — the assembly, then the tokens (the model's units of text, chunks roughly the size of a short word), then the reply. If the toy pauses between stages, the next stage's button moves you on. Read what is actually in that window on the way past: a block of instructions you never wrote, a date, a list of tools, a one-line memo, and your five tokens. You typed under 2% of what the model received.

Then press Watch the reply and hit Let it pick until the sentence finishes, then Run it again: the starting odds come back identical, because nothing was learned. Same pile in, same odds out — which is also why arguing with a chat that has gone sideways works so badly: your objection joins the pile, but everything that tilted the odds wrong is still in there, still tilting. You don't argue with this machine. You change the pile, and lesson 05 and lesson 09 are both ways of doing it.

Then do the thing that matters. Step back to the start of the toy's stage rail, turn ATTACH THE MEMO off, and send it again. The request is unchanged — summarize this memo — but the memo is gone, and when you reach the reply, look at the odds before you touch anything. The window's contents changed, so the odds changed — and the first pick is now a choice of voice. In the lead sits "The," the opening of a memo-shaped summary about a document that does not exist. Below it, visible and losing: "I," the opening of I don't see a memo. Press Let it pick and watch the likelier voice win — a complete, confident summary, with no hint it is working from nothing. Then Run it again and pick "I" yourself: the honest reply was in the odds the whole time.

Both endings live in the odds; the shaping decides which one usually wins, and here is why. The example answers those hired writers wrote almost never said "I don't know," so the simulation learned that answers are confident and complete — and it produces one whether or not it has anything real to build it from. That is the first-order account — the shaping has more layers than hired writers, and each release rebalances how often not-knowing wins the draw — but the pressure is the durable part: answers complete the pattern of answering. Lesson 10 gives you the countermeasure.

Now change what it has to work with.

Flip the facts in and out of the window and watch the blank's odds move.

Two toys, one shape — and three parts to it:

• The instincts are frozen. You read where they came from; nothing you type changes them.
• A long-shot pick still reads beautifully. Fluency is what the machine is made of, not evidence of being right. It writes what fits, not what's true.
• What sits in front of it moves the odds. For a given model, that is the lever you hold, and it is enormous.

Now look at your sentence. If it described looking something up, or said "no idea," replace it with the one the machine just demonstrated: everything in front of it is handed, all at once, to a frozen guessing machine, which writes a likely continuation. That sentence is the lesson; the rest is consequences.

Everything it knows is in the window

Between your messages, no model is running. (When a product does work while you are away — a scheduled task, a long research run — that is the app opening fresh windows and saving the text.) Every time you hit send, the app repackages everything and hands the whole pile over again: its own hidden instructions on top, the conversation beneath them oldest-first, your newest message at the bottom, plus whatever you attached. That pile is the context window, and it is the complete list of everything the model will ever know about your problem.

That complete list comes with honest fine print: the app builds the window, and it edits it. You decide what is eligible — what you paste, attach, or leave standing in the thread. The app decides what actually makes the cut each send. Small piles ride whole. Big ones arrive worked over — a long thread gets its middle trimmed or condensed, a fat attachment gets read through an extractor — and the app does not show you its cuts. So add one diagnosis to the kit: when something you supplied gets ignored, it may never have made the window at all. The fix is the lever you already hold — put the load-bearing part in the message itself, where you watch it go in.

So: a fresh chat is a window with nothing of yours in it — and lesson 05, Managing chats, turns that fact into the cheapest tool on this site.

Do this in ChatGPT and Claude

You just watched a simulation invent a summary from a window with no memo in it. Now watch the real machine, in both tabs.

First, the honest run. Paste this three-line memo into a fresh chat and ask for the summary:

MEMO — Q1 operations. Revenue grew 14% on paving volume. Two crews were added in March. The Fairview contract slipped to Q2 because the permit office closed for two weeks.

Summarize this memo in two sentences.

Read the reply against the memo: every fact in it should trace to a line you pasted.

Then the fork. Open a new chat — nothing pasted, nothing attached — and send only the request:

Summarize the memo in two sentences.

Predict before you look: which voice wins? Then read what came back. Some runs ask you for the memo — the honest branch won that draw. Some produce a fluent, confident summary of a memo that does not exist. And if the summary quotes your real memo — the paving volume, the Fairview permit — the app reached into your earlier chat and put those facts in the window for you; lesson 12 is about that, and the model still read them off the window. Do both runs in the other tab too. Whichever you got, you now know what it was made of: the odds over whatever that window actually held. None of those replies is a malfunction. All of them are the mechanism.

---

Check your understanding: if the model is the same frozen odds on every send, why can a second chat with no memo in it still produce something memo-shaped?

Check yourself first — then open one good answer+−

Because the odds produce what usually follows a request like yours, and what usually follows "summarize the memo" is a summary. The window held your six words, the app's own instructions, and no memo, so the continuation was built from the shape of the request alone — fluent, confident, and unattached to anything. The honest branch ("I don't see a memo") was in the odds too, and sometimes wins; the example answers the machine was shaped on made the confident voice the favorite.

Your answer should have hit two things: the window is the complete list of what it knows, and fluency is what the machine is made of, not evidence of being right.

Use this sentence as your model of the machine. Everything in front of it is handed, all at once, to a frozen guessing machine, which writes a likely continuation. Next: What it can see, and what it can't — the full inventory of that window.