“My name is Ozymandias, King of Kings; Look on my Works, ye Mighty, and despair!” — Percy Bysshe Shelley, Ozymandias
“The first ultraintelligent machine is the last invention that man need ever make.” — I. J. Good, 1965
1. The question
In Intelligence Is Adaptive Control of Energy Through Information, I argued that intelligence is adaptive control of energy through information: the degree to which a system can use information and feedback to capture, store, allocate, and recruit energy and other resources into goal-relevant work, while preserving itself or its goals and maintaining or expanding future options across changing conditions.
This essay is a civilizational application of that lens.
If intelligence is what lets a system steer energy adaptively rather than merely expend it blindly, then civilizations reveal a great deal about themselves by what they do with surplus. Surplus energy. Surplus labor. Surplus coordination. Surplus competence. Surplus legitimacy. Surplus state capacity.
Sometimes they build pyramids.
Or at least that is how the past looks to us.
Egypt built the Great Pyramid. The Khmer built Angkor Wat. The Inca built Coricancha, the Golden Enclosure. Mughal India built the Taj Mahal. The old world seems full of concentrated splendor: singular places where stone, gold, marble, labor, craft, and legitimacy were compressed into something undeniable.
Then we look around and ask: where is our Taj Mahal? Where is our Colossus? Where is our mountain of stone, our city of gold, our public proof that our civilization has immense surplus and wants everyone to know it?
At first glance, the comparison makes modernity look shabby. We build warehouses, data centers, chip fabs, highways, ports, container terminals, power plants, and anonymous industrial sprawl. We build things that are vastly more powerful than ancient monuments in almost any practical sense, but they rarely look like monuments. They look like infrastructure.
So it is tempting to say that the ancients knew how to build for awe, and we do not.
I think that is wrong.
2. Monuments as governance tools
The first thing to notice is that the old monuments were not really useless. They were political tools, religious tools, military tools, legitimacy tools. They were strategic signaling in stone. A pyramid says: we can command this much labor, feed this many workers, quarry and transport at this scale, and concentrate wealth on something whose main output is symbolic power. A golden temple says: this regime has tribute, craft, sacred narrative, and permanence. A cathedral says: this order binds earth to heaven.
The monument was not separate from governance.
Awe was governance.
It made power legible. It told subjects, neighbors, merchants, priests, and enemies that this civilization was real, this order was stable, and this ruler could reach far beyond the ordinary scale of human effort. That is one reason ancient monuments still grip us. They were built to.
But the deeper mistake is thinking the contrast is beauty versus function. The real contrast is that ancient civilizations concentrated splendor, while modern civilization distributes capability.
The old world often looked magnificent because it put magnificence in a few places. The pyramid was the beacon. The palace was the center. The sacred precinct was where geometry, stone, labor, wealth, and narrative came together into something undeniable. Outside those peaks, life for most people was harsh: no reliable sanitation, no climate control, no cheap glass, no artificial light on demand, no refrigeration, no antibiotics, no stable indoor comfort, no smooth roads over large territories, and no baseline material life remotely close to what even many ordinary modern people take for granted.
Ancient civilization often looked more splendid because it left so much of ordinary life in squalor. Modern civilization often looks more drab because it did the reverse. We took a large share of what would once have been concentrated splendor and spread it into the floor of ordinary life. We each get some marble now, not all of it buried with the king. We each get glass, dry interiors, plumbing, sewage, refrigeration, insulation, hot water, artificial light, roads that mostly work, instant communication, and access to information that would have seemed supernatural in almost any earlier century.
The old world built beacons. We built floors.
And perhaps that is for the better.
3. Our monuments are systems
This is the part that feels connected to Joy in the Merely Real. If you only look for wonder in domes, marble, sacred avenues, and gold, modern civilization can look spiritually thin. But that is partly a failure of perception.
Pipes, grids, cold chains, container ships, and semiconductor fabs: we are bad at seeing these as monuments because we live inside them. They are the wonders that disappeared into the baseline.
The ancients concentrated beauty because they had fewer ways to turn surplus into generalized power and generalized comfort. We can do more with the same civilizational surplus, so we often spend it less visibly. Not less impressively. Less visibly.
Nvidia and ASML’s semiconductor supply chains, continental electrical grids, industrial agriculture, global logistics networks, and SpaceX’s reusable rockets: these are our monuments.
The institutions behind them are part of the monument too: standards bodies, engineering schools, grid operators, foundries, shipping systems, regulators, and the tacit norms that keep the whole machine from flying apart. The difference is not that we stopped building wonders. It is that our greatest works are less often singular objects in the landscape and more often giant coordinated processes that disappear into the background precisely because they work.
4. The moon landing was the old logic translated into rockets
This is why Apollo, and especially the moon landing, still strikes people as special. It was one of the rare modern projects that still looked, from the outside, like a classical civilizational monument. A great power openly decided to spend a noticeable share of national surplus on a visible feat whose purpose was scientific, geopolitical, military-adjacent, and prestige-laden all at once.
That is not a side project. That is a civilization saying: look what we can do.
Apollo was not a break from the ancient logic.
It was the ancient logic translated into rockets.
The same is true, in a darker register, of aircraft carriers, stealth bombers, missile silos, and nuclear submarines. Future archaeologists may look at them the way we look at pyramids: gigantic concentrations of fuel, metal, knowledge, discipline, and surplus whose practical value lay partly in deterrence, signaling, and ritualized readiness.
They will not be entirely wrong.
Civilizations always turn surplus into whatever their environment rewards. The old world rewarded visible sacred-political splendor. The modern world rewards productive capability, military leverage, scientific depth, and systems that compound.
That is why our monuments look different.
And this is where the argument stops being merely literary and becomes concrete. We are still perfectly willing to spend pyramid-scale wealth. We just do not spend it on giant stone triangles. We spend it on chips, data centers, models, software, electrical capacity, cooling systems, factories, and institutions.
A pharaoh transported stone to make power visible. We route electricity into compute to make power more general.
Not from grandeur to banality, but from symbolic power to direct capability.
5. From symbolic power to material capability to intelligence itself
The story so far is that civilizations reveal themselves by what they do with surplus. Ancient civilizations often converted surplus into visible symbolic power. Modern industrial civilization became much more direct: instead of concentrating surplus into monuments, it converted surplus into distributed material capability. It raised the floor. It built grids, ports, fabs, farms, sewer systems, supply chains, and institutions. Its greatness became less visible, but more functional.
AI is the next step in that progression.
Ancient monuments converted surplus into visible symbolic power. Modern industrial systems convert surplus into distributed material capability. AI is the first automated, industrial-scale attempt to convert civilizational surplus into more intelligence itself.
That is the real threshold.
A pyramid took surplus and froze it into stone. A fab takes surplus and turns it into productive capacity. AI takes surplus and tries to turn it into additional cognition.
Or more precisely:
AI is civilization’s attempt to industrialize intelligence itself.
That is what makes it different from earlier monuments. A pyramid displayed power. A dam harnessed power. A factory multiplied power. AI aims at increasing the general faculty from which future power is built.
If the earlier essay was basically right, then intelligence is not just one more resource among others. It is the thing that lets civilization harvest, route, save, coordinate, and deploy energy across almost every other domain.
So what are we doing with frontier AI?
We are spending energy, matter, chip fabrication, cooling, networking, software, engineering time, and the accumulated scientific and cultural output of humanity to build systems that can model, predict, design, code, discover patterns, compress knowledge, and increasingly participate in the cognitive loops by which civilization steers itself.
That is a very weird thing to be doing.
It is not just another luxury good. It is not just another machine. It is not even just another infrastructure layer.
It is a civilizational project aimed at manufacturing more steering capacity.
6. Definition: AI is an energy pump
By this I do not mean that AI creates energy from nothing, escapes thermodynamics, or functions like a perpetual-motion machine. I mean that if intelligence is the faculty by which a system uses information and feedback to capture, store, allocate, and recruit energy and other resources into goal-relevant work, then a machine that lets civilization convert large flows of energy, matter, and accumulated knowledge into additional intelligence can create a positive feedback loop.
Energy powers compute. Compute produces intelligence. Intelligence improves design, extraction, prediction, automation, scientific discovery, and coordination. That greater intelligence then helps civilization capture, route, save, and deploy still more energy. And the loop runs again.
Before this, the main way to make more intelligence was to raise and train more humans. That worked astonishingly well. But it is slow, expensive, fragile, and only loosely coupled to available energy. You do not just add a few more power plants and get a few million more competent engineers twenty years later.
AI is different. It is the first serious attempt to make additional intelligence by something closer to industrial process: more chips, more power, more data, more training, more engineering, better algorithms, better feedback loops.
That is why it feels like such a deep civilizational break.
Not because it is “smart” in some vague science-fiction sense.
Because it changes the relationship between energy and intelligence.
A pyramid stores surplus. AI compounds it.
Of course, more steering capacity is not the same thing as wise or aligned steering capacity; a stronger pump can still pump in the wrong direction.
But that caveat does not change the underlying point. The pyramid proved that a civilization could move stone. AI suggests that a civilization may be learning to build more mind.
7. The aggregated inheritance of the species
Human history is the quarry for this new monument. We are not just burning electricity; we are distilling the recorded cognition of countless writers, coders, scientists, engineers, artists, and institutions into a single actionable lens.
That is why AI feels like such a strange threshold. Previous monuments displayed power. AI aims to increase the general capacity from which future power is built.
And once you see that, the horizon shifts quickly. First toward more efficient use of Earth’s energy and matter. Then toward more automation, more robotics, more better-than-human design and coordination. Then toward greater reach into near space, more ability to exploit off-world mass and solar energy, and more expansion into the wider physical resource base available to civilization.
A civilization that can turn energy into intelligence, and intelligence into still more energy-harvesting capacity, is on a very different trajectory from one that can only pile limestone into the sky.
The ancient world used surplus to display power. Modern industry used surplus to build distributed capability. We are beginning to use surplus to manufacture the faculty from which future power is built.
We are no longer just piling stone against the sky.
1. The question
In Intelligence Is Adaptive Control of Energy Through Information, I argued that intelligence is adaptive control of energy through information: the degree to which a system can use information and feedback to capture, store, allocate, and recruit energy and other resources into goal-relevant work, while preserving itself or its goals and maintaining or expanding future options across changing conditions.
This essay is a civilizational application of that lens.
If intelligence is what lets a system steer energy adaptively rather than merely expend it blindly, then civilizations reveal a great deal about themselves by what they do with surplus. Surplus energy. Surplus labor. Surplus coordination. Surplus competence. Surplus legitimacy. Surplus state capacity.
Sometimes they build pyramids.
Or at least that is how the past looks to us.
Egypt built the Great Pyramid. The Khmer built Angkor Wat. The Inca built Coricancha, the Golden Enclosure. Mughal India built the Taj Mahal. The old world seems full of concentrated splendor: singular places where stone, gold, marble, labor, craft, and legitimacy were compressed into something undeniable.
Then we look around and ask: where is our Taj Mahal? Where is our Colossus? Where is our mountain of stone, our city of gold, our public proof that our civilization has immense surplus and wants everyone to know it?
At first glance, the comparison makes modernity look shabby. We build warehouses, data centers, chip fabs, highways, ports, container terminals, power plants, and anonymous industrial sprawl. We build things that are vastly more powerful than ancient monuments in almost any practical sense, but they rarely look like monuments. They look like infrastructure.
So it is tempting to say that the ancients knew how to build for awe, and we do not.
I think that is wrong.
2. Monuments as governance tools
The first thing to notice is that the old monuments were not really useless. They were political tools, religious tools, military tools, legitimacy tools. They were strategic signaling in stone. A pyramid says: we can command this much labor, feed this many workers, quarry and transport at this scale, and concentrate wealth on something whose main output is symbolic power. A golden temple says: this regime has tribute, craft, sacred narrative, and permanence. A cathedral says: this order binds earth to heaven.
The monument was not separate from governance.
Awe was governance.
It made power legible. It told subjects, neighbors, merchants, priests, and enemies that this civilization was real, this order was stable, and this ruler could reach far beyond the ordinary scale of human effort. That is one reason ancient monuments still grip us. They were built to.
But the deeper mistake is thinking the contrast is beauty versus function. The real contrast is that ancient civilizations concentrated splendor, while modern civilization distributes capability.
The old world often looked magnificent because it put magnificence in a few places. The pyramid was the beacon. The palace was the center. The sacred precinct was where geometry, stone, labor, wealth, and narrative came together into something undeniable. Outside those peaks, life for most people was harsh: no reliable sanitation, no climate control, no cheap glass, no artificial light on demand, no refrigeration, no antibiotics, no stable indoor comfort, no smooth roads over large territories, and no baseline material life remotely close to what even many ordinary modern people take for granted.
Ancient civilization often looked more splendid because it left so much of ordinary life in squalor. Modern civilization often looks more drab because it did the reverse. We took a large share of what would once have been concentrated splendor and spread it into the floor of ordinary life. We each get some marble now, not all of it buried with the king. We each get glass, dry interiors, plumbing, sewage, refrigeration, insulation, hot water, artificial light, roads that mostly work, instant communication, and access to information that would have seemed supernatural in almost any earlier century.
The old world built beacons. We built floors.
And perhaps that is for the better.
3. Our monuments are systems
This is the part that feels connected to Joy in the Merely Real. If you only look for wonder in domes, marble, sacred avenues, and gold, modern civilization can look spiritually thin. But that is partly a failure of perception.
Pipes, grids, cold chains, container ships, and semiconductor fabs: we are bad at seeing these as monuments because we live inside them. They are the wonders that disappeared into the baseline.
The ancients concentrated beauty because they had fewer ways to turn surplus into generalized power and generalized comfort. We can do more with the same civilizational surplus, so we often spend it less visibly. Not less impressively. Less visibly.
Nvidia and ASML’s semiconductor supply chains, continental electrical grids, industrial agriculture, global logistics networks, and SpaceX’s reusable rockets: these are our monuments.
The institutions behind them are part of the monument too: standards bodies, engineering schools, grid operators, foundries, shipping systems, regulators, and the tacit norms that keep the whole machine from flying apart. The difference is not that we stopped building wonders. It is that our greatest works are less often singular objects in the landscape and more often giant coordinated processes that disappear into the background precisely because they work.
4. The moon landing was the old logic translated into rockets
This is why Apollo, and especially the moon landing, still strikes people as special. It was one of the rare modern projects that still looked, from the outside, like a classical civilizational monument. A great power openly decided to spend a noticeable share of national surplus on a visible feat whose purpose was scientific, geopolitical, military-adjacent, and prestige-laden all at once.
That is not a side project. That is a civilization saying: look what we can do.
Apollo was not a break from the ancient logic.
It was the ancient logic translated into rockets.
The same is true, in a darker register, of aircraft carriers, stealth bombers, missile silos, and nuclear submarines. Future archaeologists may look at them the way we look at pyramids: gigantic concentrations of fuel, metal, knowledge, discipline, and surplus whose practical value lay partly in deterrence, signaling, and ritualized readiness.
They will not be entirely wrong.
Civilizations always turn surplus into whatever their environment rewards. The old world rewarded visible sacred-political splendor. The modern world rewards productive capability, military leverage, scientific depth, and systems that compound.
That is why our monuments look different.
And this is where the argument stops being merely literary and becomes concrete. We are still perfectly willing to spend pyramid-scale wealth. We just do not spend it on giant stone triangles. We spend it on chips, data centers, models, software, electrical capacity, cooling systems, factories, and institutions.
A pharaoh transported stone to make power visible. We route electricity into compute to make power more general.
Not from grandeur to banality, but from symbolic power to direct capability.
5. From symbolic power to material capability to intelligence itself
The story so far is that civilizations reveal themselves by what they do with surplus. Ancient civilizations often converted surplus into visible symbolic power. Modern industrial civilization became much more direct: instead of concentrating surplus into monuments, it converted surplus into distributed material capability. It raised the floor. It built grids, ports, fabs, farms, sewer systems, supply chains, and institutions. Its greatness became less visible, but more functional.
AI is the next step in that progression.
Ancient monuments converted surplus into visible symbolic power.
Modern industrial systems convert surplus into distributed material capability.
AI is the first automated, industrial-scale attempt to convert civilizational surplus into more intelligence itself.
That is the real threshold.
A pyramid took surplus and froze it into stone. A fab takes surplus and turns it into productive capacity. AI takes surplus and tries to turn it into additional cognition.
Or more precisely:
AI is civilization’s attempt to industrialize intelligence itself.
That is what makes it different from earlier monuments. A pyramid displayed power. A dam harnessed power. A factory multiplied power. AI aims at increasing the general faculty from which future power is built.
If the earlier essay was basically right, then intelligence is not just one more resource among others. It is the thing that lets civilization harvest, route, save, coordinate, and deploy energy across almost every other domain.
So what are we doing with frontier AI?
We are spending energy, matter, chip fabrication, cooling, networking, software, engineering time, and the accumulated scientific and cultural output of humanity to build systems that can model, predict, design, code, discover patterns, compress knowledge, and increasingly participate in the cognitive loops by which civilization steers itself.
That is a very weird thing to be doing.
It is not just another luxury good. It is not just another machine. It is not even just another infrastructure layer.
It is a civilizational project aimed at manufacturing more steering capacity.
6. Definition: AI is an energy pump
By this I do not mean that AI creates energy from nothing, escapes thermodynamics, or functions like a perpetual-motion machine. I mean that if intelligence is the faculty by which a system uses information and feedback to capture, store, allocate, and recruit energy and other resources into goal-relevant work, then a machine that lets civilization convert large flows of energy, matter, and accumulated knowledge into additional intelligence can create a positive feedback loop.
Energy powers compute. Compute produces intelligence. Intelligence improves design, extraction, prediction, automation, scientific discovery, and coordination. That greater intelligence then helps civilization capture, route, save, and deploy still more energy. And the loop runs again.
Before this, the main way to make more intelligence was to raise and train more humans. That worked astonishingly well. But it is slow, expensive, fragile, and only loosely coupled to available energy. You do not just add a few more power plants and get a few million more competent engineers twenty years later.
AI is different. It is the first serious attempt to make additional intelligence by something closer to industrial process: more chips, more power, more data, more training, more engineering, better algorithms, better feedback loops.
That is why it feels like such a deep civilizational break.
Not because it is “smart” in some vague science-fiction sense.
Because it changes the relationship between energy and intelligence.
A pyramid stores surplus. AI compounds it.
Of course, more steering capacity is not the same thing as wise or aligned steering capacity; a stronger pump can still pump in the wrong direction.
But that caveat does not change the underlying point. The pyramid proved that a civilization could move stone. AI suggests that a civilization may be learning to build more mind.
7. The aggregated inheritance of the species
Human history is the quarry for this new monument. We are not just burning electricity; we are distilling the recorded cognition of countless writers, coders, scientists, engineers, artists, and institutions into a single actionable lens.
That is why AI feels like such a strange threshold. Previous monuments displayed power. AI aims to increase the general capacity from which future power is built.
And once you see that, the horizon shifts quickly. First toward more efficient use of Earth’s energy and matter. Then toward more automation, more robotics, more better-than-human design and coordination. Then toward greater reach into near space, more ability to exploit off-world mass and solar energy, and more expansion into the wider physical resource base available to civilization.
A civilization that can turn energy into intelligence, and intelligence into still more energy-harvesting capacity, is on a very different trajectory from one that can only pile limestone into the sky.
The ancient world used surplus to display power. Modern industry used surplus to build distributed capability. We are beginning to use surplus to manufacture the faculty from which future power is built.
We are no longer just piling stone against the sky.
We are trying to teach the sky how to think.