A wooden direction pole with chapter labels, Adam and Prof. Nova standing beside it.

Start

How LLMs work, Explained Like You’re 5

You already talk to ChatGPT. These pages walk the ideas that made that possible — one research paper at a time, in everyday language, with pictures.

Start here

How does a computer know similar words?

Words used to be nametags. Then they got neighborhoods.

Because a neighborhood is not a sentence

How does AI figure out which words matter?

Every word looks at every other word — all at once.

Because looking around still needs practice

What if we just keep guessing the next word?

Books as practice. Then a little extra training for a real job.

Because guessing ahead only looks backward

Why not look at the words on both sides?

Hide a word in the middle. Peek left and right to fill it in.

Because a new job still meant extra training

What if a few examples in the chat were enough?

Make the guesser huge. Show it the pattern. Don’t retrain.

Because a giant guesser still isn’t a helper

Why doesn’t it just do what I asked?

People show good answers, then rank the rest.

This page was crafted with the help of Cursor and ChatGPT. Learn more on GitHub. Follow me on X @riseandshaheen