It’s drizzling in Shanghai today, only around 11 or 12 degrees. This kind of weather is perfect for curling up at home and casually chatting about something. Coincidentally, a reader asked me a couple of days ago why humans are so obsessed with creating artificial intelligence. Do we genuinely want to give machines a soul?
I was chatting about this topic with a friend at a bar once, and we reached a consensus: Humans actually don’t care at all whether machines have souls; we just love to over-imagine things.
The “Unrivaled Master” Hidden in a Cabinet
You might not know that the first time humanity was profoundly shocked by “artificial intelligence” was actually nearly 300 years ago.
In 1770, a Hungarian inventor named Wolfgang von Kempelen created a contraption called the “Mechanical Turk.” You can think of it as an oversized wooden arcade machine. A mannequin dressed in Ottoman robes and a turban sat at a desk with a chessboard on it.
Before every performance, Kempelen would dramatically open the cabinet doors to show the audience—it was densely packed with gears, springs, and complex mechanical structures. Then he would take out a key, wind it up, and the wooden figure would click and clack to life, playing chess against audience members.
The whole of Europe was astounded. This thing not only played chess but played it exceptionally well. It toured with its master for decades and even traveled to the United States later on. Even Benjamin Franklin played a match against it in Paris—and lost. Everyone thought, “My goodness, humanity has actually created a machine with intelligence!”
But the real truth? It was an outrageously deceptive magic trick.
Those gears in the cabinet were just a smokescreen. Behind that complex machinery actually hid a living person—usually a petite chess master. Stifled in that hot, cramped hidden compartment, lighting a candle, they silently manipulated the wooden mannequin’s arms from below using magnets and levers.
Look at this cabinet full of gears; it was all just a cover for the man hiding inside
It was just a wooden box without even electricity, yet it fooled humanity for a full 84 years. Honestly, if people only cared about “technology,” they would have spotted the flaws much earlier. But in an era before the lightbulb was even invented, people would rather believe a wooden dummy was endowed with wisdom by the Creator than believe a sweating middle-aged man was hiding inside.
When a Secretary Kicked the Programmer Out of the Room
If people in the 18th century were fooled because they didn’t understand science, then by the 20th century, when people who actually understood science tripped up, the scene was even more hilarious.
In 1966, an MIT computer scientist named Joseph Weizenbaum wrote the world’s first chatbot program, called ELIZA.
This ELIZA was incredibly dumb. It had no real understanding capability; it completely relied on simple word substitutions. For example, if you typed “I am very sad today,” it would grab the word “sad” and apply a template to ask you back, “Why do you feel sad?” If you typed “My mother is not nice to me,” it would reply, “Tell me more about your mother.”
That was the extent of its abilities. Weizenbaum himself knew perfectly well that it was just a simple text game without a single ounce of so-called “intelligence.”
But he never expected that the first person to fall for it would be his own secretary.
This secretary watched Weizenbaum write this code every day; she knew better than anyone that behind the screen were just a few lines of emotionless, broken code. However, one day, after sitting at the computer and chatting with ELIZA for a while, the secretary suddenly turned to Weizenbaum and said, “Boss, could you please leave the room for a moment? I want to chat with it privately.”
Weizenbaum was absolutely dumbfounded.
This chin-resting photo is Weizenbaum, probably despairing over human intelligence
He later roasted the incident in an article, his tone full of disbelief. He discovered that humans have a terrifying instinct: As long as a machine shows even the slightest human-like feedback, humans will frantically project emotions onto it. Later, the field of psychology simply named this phenomenon the “ELIZA effect.”
This big shot, who originally had a promising future in the computer field, was deeply traumatized by this incident. For the second half of his life, he completely transitioned into a staunch AI opponent, spending his days writing books to warn people not to trust computers too much. It’s quite a tough break—he originally just wanted to write a quick program for practice, but ended up forcing himself to become the tech world’s whistleblower.
We Were One Step Away from the First AI Being a Bartender
Speaking of ELIZA, there is actually another super fascinating piece of trivia.
Because ELIZA was originally programmed to mimic the tone of a psychotherapist—the kind of style where no matter what you say, it slowly and methodically asks follow-up questions based on your words—people spent the next few decades assuming that the initial form of AI should be a wise, calm sage who could help you solve your problems.
But when I was doing research earlier, I came across an interview with Weizenbaum that completely shattered this filter.
A reporter asked him why he chose to set the AI as a psychotherapist and whether there were any profound academic considerations. This honest guy frankly replied that he just needed to find a profession where “even if you don’t understand anything, you can still confidently keep the conversation going,” so he conveniently chose a psychotherapist. He admitted that if he had thought about it for another 10 minutes, he probably would have set it as a bartender.
It’s hilarious to think about. If he had actually thought about it for another 10 minutes, our inherent impression of AI today might be a seasoned smooth-talker wiping glasses while listening to you complain about your life. An unintentional decision drastically raised the psychological expectations of future generations toward chatbots.
Actually, Weizenbaum had seen through humanity’s true colors early on. Naming this program ELIZA wasn’t a random choice. The name was taken from Eliza Doolittle, the flower girl in the play Pygmalion. And the inspiration for this play comes from the Greek myth of Pygmalion—the king who carved a perfect stone statue of a maiden and eventually fell hopelessly in love with his own creation.
Take a moment to savor this naming. When we build artificial intelligence, a lot of times we are just like that king hugging and kissing a stone. We are not obsessed with some great soul being born inside the machine; we are merely looking into this cold mirror, seeing the reflections of our own displaced emotions.
Who Cares If It’s Real or Not
Many times, it really isn’t because we eagerly crave witnessing the birth of a brand-new silicon-based life. The truth might be far more mundane: On this planet crowded with billions of people, we still feel incredibly lonely.
So much so that whether the entity on the other side is an old man playing chess hidden in a wooden cabinet or a few lines of clumsy substitution code that doesn’t even have a memory function, as long as it is willing to quietly listen to us finish speaking at a specific moment—without interrupting, arguing, or judging—we are more than willing to believe it has a soul.
It’s like shouting into a hollow tree; you don’t really care if there’s a fairy inside, you just need an echo.
The rain outside seems to be getting heavier, and I’m hungry after talking for so long. I need to go order some takeout and perhaps pick a fight with my smart speaker along the way. Have you guys ever lost your temper at something you fully knew was just a machine?
References：
- The Mechanical Turk: AI Marvel or Parlor Trick? – Britannica
- Mechanical Turk – Wikipedia
- Why Joseph Weizenbaum Invented the Eliza Chatbot – Smithsonian Magazine
- Weizenbaum’s nightmares: how the inventor of the first chatbot turned against AI – The Guardian
—— Lyra Celest @ Turbulence τ.
