Chapter Three - The First Boom

CHAPTER THREE

The First Boom

THERE WAS A time when artificial intelligence wasn’t a maybe — it was a soon. A guaranteed future. A solved problem. Just a few more lines of code, a few more circuits, and we’d have thinking machines that could match or surpass us.

Welcome to the AI gold rush of the 1960s — a strange cocktail of math, hubris, government grants, and absolute faith that intelligence was just a logic tree waiting to be mapped.

Spoiler: it didn’t end well.

Let’s start with the perceptron.

Invented by Frank Rosenblatt in 1958, the perceptron was an early artificial neural network — a machine that could learn to categorize things by adjusting internal weights based on examples. Feed it enough images, and it could learn to distinguish cats from dogs. Or missiles from birds. Or faces from not-faces.

The Pentagon loved it. The press went nuts. Newspapers declared that the perceptron would soon walk, talk, and reproduce.

They were wrong. But we’ll get to that.

Meanwhile, a guy named John McCarthy — who actually coined the term artificial intelligence — was developing a programming language called Lisp. It wasn’t just a way to write code. It was a way to simulate reasoning itself.

Lisp became the go-to tool for AI research. It was flexible, recursive, and symbolic — perfect for building systems that could manipulate ideas rather than just numbers. If you wanted your machine to understand language, solve puzzles, or play games, you probably built it in Lisp.

And for a while, it worked.

Then there was Marvin Minsky — the face of AI for decades.

Minsky believed that the mind could be broken down into modules, and that machines could replicate those modules. He saw no reason why we couldn’t build artificial brains with the right architecture. He once said: “In from three to eight years, we will have a machine with the general intelligence of an average human being.”

That was in 1970.

But at the time, no one laughed. The U.S. government was pouring money into AI. The Defense Department wanted robots that could translate Russian, interpret satellite images, and command battlefields. The dream was alive.

And the grants were flowing.

In 1966, a programmer named Joseph Weizenbaum created ELIZA — a chatbot that mimicked a Rogerian psychotherapist. It asked vague questions like “How does that make you feel?” and “Tell me more about your mother.”

It was a gimmick. A trick. A glorified Mad Libs program.

But people believed it. Some users insisted ELIZA understood them. One secretary asked to be left alone with the program, believing it truly empathized.

This was the first glimpse of a terrifying truth: people want to believe. Even fake minds can have real power.

But beneath the headlines and hype, something was wrong.

The perceptron couldn’t do basic math. Symbolic AI struggled with real-world ambiguity. Language understanding hit hard limits. The moment you took these systems outside the lab, they fell apart.

AI wasn’t learning. It was bluffing.

And soon, the funding would dry up.