Free preview6 min

The Autocomplete That Ate the Internet

Reviewed by Human · Updated July 13, 2026

You've heard that AI 'thinks'. Myth. The most powerful AI on the planet does exactly one thing: it guesses the next word.

 

That's the whole trick. Not a brain. Not a mind. A guessing machine — trained on more text than you could read in a thousand lifetimes. And once you see the trick, you'll never un-see it.

Real world

Your phone keyboard. You type 'See you' and it suggests 'later', 'soon', 'tomorrow'. It's not reading your mind. It has seen millions of texts, and those words usually come next.

AI parallel

A large language model is that same suggestion bar — scaled up enormously. It reads everything you gave it, then picks the most likely next word. Then the next. Then the next. Whole essays, one guess at a time.

Live Demo — AI Terminal

Try this with any AI chat. Type an unfinished sentence and nothing else. Watch what happens.

Your prompt

How Does AI Predict the Next Word?

Time for three terms you'll use forever. **LLM**: A large language model — an AI trained on huge amounts of text to predict what comes next. **Token**: A small chunk of text, roughly a word or word-piece, that the model reads and produces one at a time. **Next-token prediction**: The single task an LLM performs — picking the most likely next token, over and over, until it stops. That's the whole engine. Scale makes it shocking. GPT-3, for example, used 175 billion parameters — the internal dials tuned during training (source: arxiv.org/abs/2005.14165). Enough dials, enough text, and 'guess the next word' starts producing essays, code, and jokes.

Every impressive AI output you've ever seen was built one token at a time. No plan. No draft. Just a chain of very good guesses.

Your turn

Be the autocomplete. Finish these three sentences the way you think an AI would — go with the most obvious continuation, not the most creative one.

Reflect

You just did next-token prediction. Notice how easy it was? You've absorbed patterns from years of reading. The model did the same thing — with the whole internet.

Plot twist

▎

📚

Go deeper

end of module

You've finished this module.

Mark it complete to earn your XP and keep your streak alive.

Progress saved locally · Sign up to earn XP

Why It Sounds So Smart (But Isn't)→