Temples of Illusion: When the Ancient Middle East Invented the First Thinking Machines

There was a time when the gods spoke. In the temples of Thebes and Memphis, statues inclined their heads to designate the future pharaoh. In Babylon, superhuman guardians watched over sacred forests. In Alexandria, sanctuary doors opened on their own as worshippers approached. Pilgrims marveled, prostrated themselves, believed. They could not imagine that behind each miracle lay a mechanism, that behind each oracle worked an engineer.

The fertile crescent of the ancient Middle East—from Mesopotamia to Egypt, from the banks of the Tigris to the quays of Alexandria—was not only the cradle of writing and agriculture. It was also the first forge where humanity attempted to create the illusion of thought.

Yesterday — The Dawn of Algorithms

Nearly four thousand years ago, long before the word "calculation" existed, Babylonian scribes were engraving in clay the first systematic procedures for solving mathematical problems. These cuneiform tablets, deciphered by computer scientist Donald Knuth in 1972, reveal a strikingly modern approach: rather than recording isolated solutions, the Babylonians set down sequences of instructions applicable to entire classes of problems. They were working, in Knuth's words, with a kind of "machine language" of formulas—an algorithm, before the letter.

The sexagesimal system they invented—that base sixty which still gives us our minutes and seconds, our degrees and hours—was not a numerical caprice. It was a calculation tool of formidable efficiency, enabling divisions and multiplications that our base ten makes laborious. The famous Plimpton 322 tablet, with its Pythagorean triples too numerous and precise to have been found by trial and error, testifies to a mathematical sophistication unsuspected until recently.

More remarkable still: a discovery published in the journal Science in 2016 revealed that Babylonian astronomers, between 350 and 50 BCE, calculated Jupiter's trajectory using the area of a trapezoid representing its velocity over time. They had invented, fourteen centuries before medieval Europe, the beginnings of integral calculus. The geometry they wielded was not that of fields and cadastres. It was a geometry of abstraction, where figures represented not physical spaces but relationships between variable quantities—velocity against time, movement against duration.

These mathematicians did not build machines. But they laid the conceptual foundations without which no thinking machine could ever exist: the idea that a problem can be decomposed into steps, that a procedure can be generalized, that a calculation can be reproduced mechanically by anyone who follows the instructions.

Meanwhile, Mesopotamian mythology explored another dimension of artificial creation. The Epic of Gilgamesh, the oldest literary work of humanity, features Humbaba, guardian of the Cedar Forest. A creature with the breath of death and a gaze of fire, appointed by the god Enlil to protect a sacred domain, Humbaba is neither human nor truly divine. He is an agent, programmed for a single task: to watch, detect, destroy intruders. Some see in him today the first automaton in world literature.

And then there is Enkidu, the wild man fashioned by the gods to temper Gilgamesh's pride. Created from clay by the goddess Aruru, Enkidu was not born. He was manufactured—designed to fulfill a precise function in the order of the world. Mythology was already posing, in the language of myth, the question that still haunts us: can one create a being that thinks, feels, acts of its own accord?

Today — The Programmer-Priests

If the Babylonians invented the algorithm, the Egyptians invented the interface. As early as 2500 BCE, temples along the Nile housed statues capable of movement and speech. Egyptologist Gaston Maspero described these images of the gods, in painted or gilded wood, endowed with "articulated limbs and voices," which "answered questions and sometimes gave long speeches."

The secret lay in the engineering of concealment. Systems of levers and counterweights, housed in the hollow pedestals of statues, allowed an arm to move, a head to incline. Hidden tubes, snaking from the back of the sanctuary to the mouth or ear of the idol, transmitted the modified voice of the priests. In 1936, archaeologist Loukianoff discovered a bust of the god Re-Harmakhis pierced by a secret cavity at the back of the neck, connected by a narrow channel to an invisible orifice beneath the right ear. Anyone who spoke into the cavity would see their voice transformed, amplified, made unrecognizable—divine.

These devices were not mere technical curiosities. They constituted instruments of power. Statues "chose" the next sovereign from among the heirs by inclining their heads toward the elect. They prophesied harvests, wars, epidemics. The faithful, knowing nothing of the hidden mechanisms, attributed superhuman knowledge to their masters. The priest had become a programmer—not of code, but of belief.

The Colossi of Memnon illustrate this porous boundary between the natural and the artificial. These two giant statues of Pharaoh Amenhotep III, sixty feet tall, began to "sing" at sunrise after an earthquake cracked one of them in 27 BCE. The Greeks saw an oracle in it, naming it after the Ethiopian king of their mythology. Centuries of pilgrims came to consult the singing statue. Was it a natural phenomenon—the evaporation of dew in the porous stone, the differential expansion from morning warming? Was it manipulation by the site's guardians? The question remains open. But when the Roman emperor Septimius Severus had the statues restored in the third century, filling the cracks with stone and metal, the singing ceased forever. The oracle fell silent, and pilgrims stopped coming.

It was in Alexandria, crossroads of Egypt and Greece, that these traditions converged to give birth to a true science of automation. Ctesibius, considered the father of pneumatics, built in the third century BCE water clocks of precision unmatched for two millennia, an organ powered by hydraulic pressure—the first keyboard instrument in history—and countless entertainment devices. His student Philo of Byzantium went further still: he created a life-sized mechanical servant capable of automatically pouring wine and then water into a visitor's cup. The mechanism relied on airtight containers, air tubes, and a subtle play of gravity and pressure. The servant had no consciousness, but she had behavior—a sequence of actions triggered by an external stimulus.

Hero of Alexandria, in the first century CE, brought this art to its peak. His aeolipile, a hollow sphere set spinning by steam jets, constitutes the first heat engine in history. His temple doors, actuated by the heat of an altar fire, opened and closed without human intervention. His holy water dispenser, triggered by inserting a coin, is humanity's first automatic transaction machine. And his mechanical theater, a ten-minute spectacle entirely programmed by a system of ropes, knots, and cams, strikingly foreshadows the cam-driven automata of eighteenth-century Europe.

Alexandria became, according to historians, "the technological center of the Mediterranean world, the city of gears and cogs, of blown air and running water." In the absence of electricity, pneumatics and hydraulics had to serve in its place. And serve they did—not only to amaze the crowds, but to pose a question we still ask ourselves.

Beyond — The Question of the Temples

Where does mechanism end, where does mind begin? Egyptian priests would doubtless have answered that the boundary does not exist—that the god inhabits the statue once the appropriate rituals have been performed, that mechanics is the vehicle of the divine, not its adversary. Alexandrian engineers might have shrugged: for them, the question was less metaphysical than practical. What matters is that the device works, that the effect is produced, that wonder occurs.

We have inherited both traditions. Our artificial intelligence systems accomplish tasks we would once have deemed impossible without consciousness: they translate, they diagnose, they create. They do so through procedures—algorithms—that descend directly from Babylonian tablets. And we present them to the world through interfaces designed to produce an effect, to elicit wonder, to make us forget the mechanism working behind the scenes.

The priests of Memphis were not charlatans. They probably believed that their statues, animated by rites, became true receptacles of divine presence. The mechanism did not negate the miracle; it made it possible. Likewise, today's creators are not necessarily deceived by their creations. They know their models do not think in the way we think. But they also know that something happens in this encounter between algorithm and user—something that resembles intelligence, that produces value, that transforms the world.

The difference, perhaps, lies in scale. The speaking statues of Egyptian temples impressed a few hundred pilgrims. Today's algorithms touch billions of human beings every day. The question of illusion and reality, of mechanism and mind, is no longer a philosophical curiosity. It has become a civilizational stake.

The Babylonians taught us that calculation can be proceduralized, that intelligence can be broken down into reproducible steps. The Egyptians showed us that interface matters as much as mechanism, that the user's belief is part of the system. The Alexandrians bequeathed us the principles of automation, the conversion of energy into motion, of stimulus into response.

We are the heirs of these temples where gods spoke without being divine, where machines acted without being alive.

The next time you ask a question to a virtual assistant and an answer springs forth, seemingly from nowhere, think of the priests hidden behind the statues of Thebes. Think of the Babylonian scribes aligning their instructions on clay tablets. Think of Hero adjusting the steam jets of his aeolipile in the workshops of Alexandria.

Four thousand years separate us from them. And yet, we still ask the same question: does what answers me actually think?

The answer, perhaps, matters less than continuing to ask it.