October 20, 2022   6 mins

In 1770, the Habsburg Empress Maria Theresa and her court were astonished by a true marvel of the modern age of engineering: a humanoid machine able to defeat a human opponent at chess.

The device consisted of a life-size figure, dressed in the “Oriental” style and seated in front of a chessboard. When it defeated several opponents at court, it was a sensation: widely known as the “Mechanical Turk”, it toured France, Britain and the United States, in the course of which it played many games including against Napoleon and Benjamin Franklin.

The only hitch: the Mechanical Turk was a fake. Though the intricacy of the hoax was itself a feat of remarkable engineering, the chess-playing intelligence was supplied by a human cleverly concealed inside the “machine”.

Today, humanoid robots are so life-like they’re again triggering accusations of fakery – and, as the Mechanical Turk did, fears that intelligent machines might be poised to take over the world. When the humanoid robot Ai-Da “addressed” the House of Lords to give evidence on the impact of technology in the creative industries, some Lords were reportedly “terrified“.

But are such devices really a threat? The answer is probably not – at least, not directly. Lifelike automata are not the real threat, so much as a diversion from it.

We’ve been building mechanised imitations of humans, animals and even gods all the way back to ancient times. The ancient Greeks used automata in temples and processions: ancient sources describe mechanical figures that danced or poured libations, trumpets that sounded automatically on the opening of temple doors, and other contrivances designed to inspire awe or reverence.

This knowledge thrived in the Roman empire — but after its fall, the West largely lost interest in technology. For the medieval peoples of Christian Europe, the world was a place not of engineering challenges but mystery and enchantment: a book written by God, in which divine order suffused everything. Meddling in God’s domain was, to put it mildly, frowned upon.

By the Renaissance, the printing press was whittling away at Church monopoly on access to knowledge, while the knowledge of the ancients began to spread once more. Increasingly, the world seemed less God’s mystery than a place full of puzzles to be mastered.

Leonardo da Vinci was one of the first influential creators of automata in the creative ferment of the early modern era. Among his designs were a humanoid “knight” figure, and (reportedly) a mechanical lion whose chest opened to present a bunch of lilies to the King of France.

RenĂ© Descartes tipped the scales still further away from mystery toward engineering a century later. He advanced the idea that instead of being governed by God’s constant imposition of divine order, it worked according to mechanistic laws.

Descartes wasn’t an atheist, but in his view, God was something akin to a clockmaker, and the cosmos God’s clock. Accordingly, Descartes declared that he saw “no difference between machines built by artisans and objects created by nature alone”. Only the human soul was exempt from these mechanistic laws: the one divine spark, piloting a meat suit from a seat (Descartes hypothesised) in the pineal gland.

The mechanistic, Cartesian view of the cosmos paved the way for a frenzy of investigation, exploration and experimentation, as inventors set about seeing if they could replicate the work of the divine clockmaker. One effect of this was a flourishing golden age of automata, through the 18th and 19th centuries. Along with the Turk — a feat of engineering even if its intelligence wasn’t real —surviving examples from this period include a humanoid automaton who writes letters, another who draws pictures, and one that plays the organ.

Another marvel, later destroyed in a fire, was a mechanical duck that first wowed the French court in 1740 by flapping its wings, quacking, splashing in water and even (apparently) defecating. Onlookers included Voltaire; the court marvelled at the mechanical recreation of life, and hailed its creator as the new Prometheus.

The inventor, Jacques de Vaucanson, typified this remarkable age of mad, whimsical, intricate inventiveness — as well as the fact that those who created such machines often faced the fury of others for whom the world was still God’s book. As a novice monk, he devised a machine that served dinner – only to find that it wasn’t celebrated by his fellow-monks but denounced as diabolical. Devastated, de Vaucanson abandoned his novitiate and fled to Paris, where he made several more automata including the duck, and a “flautist” who played a real tune.

But these were all, in the end, just machines: intricate but inert. It’s in the human “ghost” in the Mechanical Turk’s “machine” that we find a clue to the shadow cast by such automatons – as well as in the other robots that de Vaucanson created. For the playful, magical and unnerving automata created to amuse kings and nobles were also the exuberant bow-wave of technologies with far more pragmatic and far-reaching consequences. The real work went on in the mechanisation of more ordinary areas of life – especially further down the social scale. And some of the complex interconnections between automation and the transformation of life for the poorest are evident in the history of the word “Robot” itself.

The modern world uses “robot” to refer to mechanical devices that carry out work previously done by humans; the origin of the term lies in the Robot: forced labour owed by the peasantry to their aristocratic landlords in the Habsburg territories.

In his history of the Habsburg empire, AJP Taylor notes that it was abolishing the Robot that began Hungary on the journey to embracing mechanisation. Taylor points out that the main effect of emancipating the peasantry was enabling the industrialisation of farming: “The great estates, freed from the inefficient Robot, could be conducted more economically. The steam ploughs of Hungary, a striking feature of the late nineteenth century, were the result of peasant emancipation.”

And these innovations were byproducts of the same frenzy of robot-inventing creativity as the automata. Along with marvels such as the Digesting Duck, in 1745 de Vaucanson also invented the first automated loom: a development that would later play a crucial role in mechanising forms of work that had previously been the preserve only of humans. And what should concern us is less whether machines will become sentient (they won’t), as what the effects of ever greater mechanisation will be on humans.

Marx observes in Das Kapital (1867) that “History discloses no tragedy more horrible than the gradual extinction of the English hand-loom weavers.” Similarly, for a glimpse at what advances in humanoid robotics are actually for at ground level, look at “Quinn“. Quinn is a customer-service robot: instead of paying several human salaries, a hotelier can install a Quinn at front desks across a whole chain, backed up by just a few remote operators, able to step in if a query is too complex for the machine. Further down the scale, self-checkout machines are effectively clumsy Quinn devices, that displace the work of sensemaking onto the customer, and troubleshooting onto a skeleton staff of supervisors.

And this displacement of human skill and intelligence in turn re-orients human work to machine priorities. Marx described the way factory assembly-lines compelled human workers to adapt their movements, working speed and behaviour to the demands of the machine, rather than employing tools in the interests of work performed to a human pattern. The same obtains, too, for subsequent waves of automation.

In the 21st century, “Mechanical Turk” gives its name to an online platform that makes dull, repetitive data work affordable for buyers by outsourcing it remotely — to workers paid, as the New York Times reported in 2019, as little as $0.97 an hour. We find many such exploited human “ghosts” in the AI “machine”: the Amazon fulfilment workers optimised to breaking-point (and, notoriously, peeing in bottles) by algorithmic surveillance, for example, or the gig-economy content moderators workers struggling with PTSD from the horrific things they’ve seen.

And there are even humans, hidden in AI every bit as uncomfortably as the Mechanical Turk’s concealed operator, whose role is to make up the shortfall in the dumb “intelligence” of the machines. Take, for example, the people hired to impersonate chatbots for companies who want to look cutting-edge.

The convergence of human and machine in turn makes actual humanness into a luxury. As mechanical weaving made textiles cheap, so hand-woven fabrics are now wildly expensive – as is anything created by hand with genuine artisan skill. Likewise, as the hospitality industry has automated and de-personalised through the Covid era, “contactless” travel has made human contact a premium extra — because what people really want is to talk to a human. One hotel industry outlet describes human assistance as “the hallmark of a luxury travel experience”.

So it doesn’t really matter if automata exist that are capable of deceiving some humans (or at least Google engineers) about their humanness. The ones that get rolled out at scale don’t bother aiming for verisimilitude, and it’s these that most radically transform our lives at scale.

While we marvel (or shudder) at the simulacra which convince, and numbly tolerate the ones that don’t, every advance re-orders another wave of human work to the machine’s priorities. And every time they do so, another facet of human warmth, intelligence and skill becomes a premium extra, for the lucky few.


Mary Harrington is a contributing editor at UnHerd.

moveincircles