
AI isn't one thing. It's a 70-year-old umbrella that has covered logic programs, expert systems, statistical learning, neural networks, and now LLMs. Part 1 builds the vocabulary from the ground up.
The AI Explainer, Part 1: Foundations
AI isn't one thing. It's a 70-year-old umbrella term that has covered logic programs, expert systems, statistical learning, neural networks, and now transformer-based language models. Most of the public confusion about AI traces back to talking about the umbrella when we should be talking about whatever's actually under it.
This first part of the trilogy builds the vocabulary from the ground up. What AI actually is, and what it is not. How machine learning inverts the way programs get written. Why a neural network is a giant adjustable function, not a brain. The 2017 paper that made modern LLMs possible. And what an LLM actually does when you type into it. Five articles. Enough mental model to stop being fooled by marketing language.
The phrase 'artificial intelligence' first appeared in print on August 31, 1955, as a marketing choice. It has since covered so many different technologies that the word itself has stopped being useful.
4 min readMachine learning isn't the computer learning like a human. It's an old mathematical idea that finally got enough data and compute to work at scale.
4 min readNeural networks have nothing to do with how brains actually work. They are functions with billions of adjustable knobs, turned by an algorithm from a 1986 Nature paper.
4 min readA 2017 arXiv paper called "Attention Is All You Need" introduced the transformer architecture. Every modern AI product you've used is built on it.
4 min readA large language model is one mechanical thing: a function that predicts what comes next. Everything else is scaffolding around that one operation.
4 min read