In 1966, a 200-Line Program Fooled People into Thinking a Computer Understood Them. We're Still Falling for It.

Key Takeaways

• •ELIZA was written in MAD-SLIP by Joseph Weizenbaum at MIT in 1966 — roughly 200 lines of code
• •Used only pattern matching and keyword substitution — zero actual comprehension
• •Weizenbaum's own secretary asked for privacy to speak with the program
• •Weizenbaum coined 'the ELIZA effect' — humans projecting understanding onto machines that have none

Source: RootByte research / published estimates

Timeline

In 1966, at MIT's Artificial Intelligence Laboratory, Joseph Weizenbaum wrote a program called ELIZA. It was about 200 lines of code in a language called MAD-SLIP. It had no neural networks, no training data, no machine learning. It used pattern matching — find a keyword in the user's sentence, slot it into a template response, spit it back.

User: 'I'm feeling sad today.' ELIZA: 'Why do you think you're feeling sad?' User: 'My mother doesn't understand me.' ELIZA: 'Tell me more about your mother.'

That was the entire trick. ELIZA's most famous script, DOCTOR, mimicked a Rogerian psychotherapist — a style of therapy where the therapist reflects the patient's words back as questions. It was the perfect disguise for a program that understood nothing. Rogerian therapy already sounds like a machine repeating what you say.

The response was immediate and disturbing. Weizenbaum's secretary — a woman who knew he had programmed ELIZA, who had watched him write the code — asked him to leave the room so she could have a private conversation with it. Psychiatrists at MIT suggested ELIZA could be used for real therapy. The program had no memory between sentences. It forgot everything the moment you stopped typing.

Weizenbaum's secretary asked him to leave the room so she could speak privately with ELIZA. The program had no memory, no understanding, and no idea she was there.

Weizenbaum was horrified. He had built ELIZA as a demonstration of how shallow natural language processing was — how easy it was to create the illusion of understanding without any actual understanding. Instead, people treated it as proof that machines could think.

He named the program after Eliza Doolittle from George Bernard Shaw's Pygmalion — a flower girl taught to speak 'properly' without comprehending the social world she was imitating. The name was a deliberate warning about mistaking performance for comprehension. Almost everyone missed the irony.

Weizenbaum spent the rest of his career warning about the dangers of attributing understanding to machines. His 1976 book, Computer Power and Human Reason, argued that certain human roles — therapists, judges, teachers — should never be delegated to computers, regardless of how convincing they seem. The book was largely ignored by the AI community.

He named it after Eliza Doolittle — a character taught to mimic without understanding. The irony was the point. Nobody got it.

He coined a term for what he'd witnessed: 'the ELIZA effect.' It describes the human tendency to project genuine understanding onto systems that are simply following patterns. In 1966, the pattern was keyword matching. In 2026, the pattern is statistical prediction across trillions of tokens. The ELIZA effect hasn't weakened. If anything, it's stronger.

When ChatGPT launched in November 2022 and gained 100 million users in two months, the conversations were eerily familiar. People confided in it. They asked it for life advice. They described it as 'understanding' them. The technology was unrecognizably more sophisticated than ELIZA. The human response was identical.

Weizenbaum died in 2008, in Berlin. He never saw ChatGPT. But he predicted it — or rather, he predicted us. The problem was never whether machines could understand. The problem was always that we so desperately want them to.

How did this make you feel?

Recommended Gear

Keep Reading

History

Python Started as a Christmas Holiday Project — Now It Runs the AI Revolution

5 min read

History

The Government Lab That Accidentally Invented the Internet, GPS, Siri, and Self-Driving Cars

7 min read

History

NVIDIA Nearly Closed Its Doors After a Sega Deal Gone Wrong. Now It's Worth $5 Trillion.

7 min read