ELIZA

We built our first chatbot before most of us had seen a computer. In 1966, Joseph Weizenbaum at MIT wrote ELIZA. She looked like magic: type words in, get words back. But the trick was smaller than it seemed.

Pattern matching

ELIZA didn’t understand language. She scanned input for trigger words. If we typed mother, she spotted it. Then she pulled a stock reply. Think of a “Find/Replace” that pretends to listen. That’s pattern matching.

Scripted responses

She kept a script—basically a list of “when X, say Y.” For therapy-style chat, “I am sad” triggered something like “How long have you been sad?” The script did the heavy lifting. Without it, she stayed quiet.

Reflection

Her neatest trick was reflection. Take our words, flip them back. If we said, “I hate my job,” she answered, “Why do you hate your job?” No reasoning, just a mirror. But people leaned in because it felt personal.

Why it worked

We filled in the gaps. Our brains turned repetition into empathy. She never claimed insight. We gave it to her. That’s why ELIZA became a milestone: not because she was smart, but because we believed she was.