The Machine That Pretended to Think
What an 18th century chess automaton reveals about artificial intelligence today.
Before we can set healthy boundaries between us and AI, we should know what AI actually is. And to get a grasp on what it actually is, it’s helpful to see where it came from. So over a few posts we’re going to glance over our shoulder and take in a brief history of thinking machines.
I could start with 19th century English inventor Charles Babbage, whose design for his Analytical Engine is considered the first digital computer. But as I was studying Babbage’s work, I caught wind of another ingenious invention that is actually a better place to begin our history of machine intelligence. It inspired Babbage in his work, and it also serves as a suitable analogy for today’s AI.
It all started with a magic trick.
In the 18th century, it was common for royal courts in Europe to host lavish events to entertain guests and advertise the best of their kingdoms. In 1769, Empress Maria Theresa of the powerful Habsburg Empire hosted a magic show. One of the civil servants employed in her court, Johann Wolfgang von Kempelen, was invited to evaluate the performance. Kempelen was a writer, engineer, and inventor known for his skill in mathematics and physics. He reported to the empress that he could produce a much better show, and he was given six months to prove it.
In those days, life-like machines—known as automata back in the day—were popular. Crowds had already been wowed by contrivances that could play musical instruments or write. Kempelen was keen to harness his engineering prowess to make something truly impressive. He constructed a large wooden cabinet that housed an intricate mass of wires, gears, belts, and other apparatus. Behind the cabinet he placed the bust of a man dressed in oriental garb, and on the top of the cabinet he placed a chessboard. In 1770, Kempelen debuted what came to be called his Mechanical Turk, a contraption that would go on to entertain crowds all over Europe and the Americas and confound scientists and common folk alike for generations.
During performances, Kempelen would open two front cabinet doors one at a time to show attendees the interior workings. Then he would insert a large key to wind the machine up. A member of the audience would sit down at the cabinet and proceed to play a game of chess with the animated machine. In thirty minutes or so, the automaton would almost always emerge as the winner, clearing the board with a flourish of its arm, much to the dismay of the human players.
Word of the amazing chess-playing machine spread fast, and soon Kempelen was taking his invention on tour to France, Germany, England, and the United States. The machine bested the chess-playing talents of some notable figures, including American statesman Ben Franklin, Frederick the Great, and even Napoleon Bonaparte. Luminaries and commoners alike attempted to beat the Turk and failed. The contraption came to represent a new and uncertain era of scientific and mechanical advancement. While some were convinced the machine was truly intelligent, skeptics were sure that it was an elaborate hoax. And though the most common explanation was that it was being controlled by a human, theories varied as to exactly how. “It is quite certain that the operations of the Automaton are regulated by mind,” wrote American author Edgar Allan Poe in an 1836 essay on the topic. “The only question then is of the manner in which human agency is brought to bear.”
The mystique, of course, was part of the automaton’s success, but for its inventor, the Mechanical Turk became a burden, holding him back from further projects. Whenever he could, Kempelen relegated his famous invention to the back of his workshop. Kempelen died in 1804, but his automaton creation lived on. It was acquired by Bavarian engineer Johann Maelzel, who continued to travel with it and give performances. Maelzel, who also invented a metronome and an advanced music box, freshened up the automaton’s act, even giving it the ability to announce “check mate” in French to opponents.
But the secret of the Mechanical Turk would not be concealed indefinitely. After Maelzel’s death, the Turk was auctioned to a doctor from Baltimore, who started a club of enthusiasts who were allowed to see how the Turk really worked as paying members. But with the cat out of the bag, it didn’t take long for interest to evaporate, and the automaton was stored in a museum before being destroyed in a fire in 1854. Three years later, the son of the last owner wrote a book detailing the workings of the elaborate machine.
The true genius of the Mechanical Turk, of course, was entirely human. Kempelen was a masterful engineer. His design of a rolling seat with greased rails allowed a hidden chess player to maneuver inside the cabinet, remain concealed even with a cabinet door open, and manage the entire chess game from below the board. The nimble talisman was able to move the Turk’s arms and fingers through a detailed system of levers built into the torso. He monitored moves on the chess board above him with magnets, providing a mirror image that he reproduced on a miniature board to keep track. And all the inner workings inside the cabinet had but one job: to conceal the human chess player within and give the illusion of a working, intelligent machine.
Machines have come a long way since the days of the Mechanical Turk. We’re in the age of programmed software, voluminous code that dictates the virtual actions of computing machines. And with today’s AI, we’ve got what very much looks like thinking machines in the pure sense: machines that can think independently. We are dazzled by AI’s ability to analyze large datasets, find patterns, and use language patterns to respond to us conversationally. But inside the cabinet of AI, it’s human skill, human achievement, and human input that allows AI to do what it does.
A few more ways AI resembles the Mechanical Turk:
The chess-playing automaton was a scientific and technological marvel, even as it gave the illusion of intelligence. AI also represents remarkable achievement in science and technology, and it too offers the illusion of intelligence.
The Turk concealed a chess master. Today’s AI conceals human civilization itself: the work of millions of humans compressed into statistical patterns.
The Turk simulated a thinking person by wearing human garb and making convincing moves on the chess boards. AI uses language and image generation to simulate mind.
The mystery of the Turk’s movements amplified its power. We don’t see the inner workings of AI either. The training process, training data, algorithms, and programmed code is all hidden from users.
The real genius of the Mechanical Turk is the creativity and skill it took to create, improve, and run the machine. Similarly, the real intelligence behind AI is us.
The popularity of the Mechanical Turk had faded away by the mid 19th century, but the intrigue around the possibilities of machine intelligence was just heating up. In 1833, Mr. Babbage had begun work on his Analytical Machine, a follow-up to his earlier Difference Engine. Babbage had played the chess-playing automaton twice in 1819 and lost. But he was convinced that machines could indeed be programmed to operate without direct human involvement. In our next look back at the history of thinking machines, we’ll explore his work and the insight of his collaborator, mathematician and writer Ada Lovelace.
