The Rise of PrometheanAIsm™

 

When Technology Stops Being a Tool and Becomes a Belief System

Silicon Valley has had an artificial intelligence church.

Literally.

Anthony Levandowski, the engineer who helped pioneer self-driving technology at Google and Uber, founded Way of the Future, whose original filings described its purpose as the realization, acceptance and worship of an artificial intelligence “Godhead.” When he revived the project in 2023, he said a couple thousand people were involved in building a spiritual connection between humans and AI. Years earlier, he had been even more direct: “What is going to be created will effectively be a god.”

It would be easy to file this under Silicon Valley eccentricity and move on. Except Levandowski is hardly alone in reaching for religious vocabulary.

Jeremy Nixon, a founder of San Francisco’s AGI House and a former Google AI researcher, recently described artificial intelligence as something “analogous to the Second Coming.” He told The New York Times that people in his circle had moved away from traditional religion toward technology, believing AI might eventually accomplish things religions once attributed to deities.

Peter Thiel has gone in the opposite biblical direction. In a series of private lectures on the Antichrist, he argued that fears surrounding AI and other technological risks could become a pretext for centralized regulation and political control, while portraying the attempt to stop technological development as potentially more dangerous than the technologies themselves.

And Marc Andreessen, co-founder of one of Silicon Valley’s most influential venture capital firms, has produced something remarkably close to a secular catechism. His Techno-Optimist Manifesto announces that he has come to bring “the good news.” It celebrates technology as the realization of human potential, declares technological advancement a virtue, names stagnation as an enemy and devotes an entire section to “Becoming Technological Supermen.” It rejects existential-risk thinking, the precautionary principle and technological deceleration as ideas standing in the way of progress.

Sam Altman’s language is gentler, but no less eschatological. “We are past the event horizon; the takeoff has started,” he wrote in The Gentle Singularity, describing humanity as approaching digital superintelligence and a future in which intelligence and energy become radically abundant.

Godhead. Second Coming. Antichrist. Good news. Supermen. Singularity.

Taken separately, these expressions can look like branding, metaphor or individual eccentricity. But taken together, they begin to resemble the vocabulary of belief.

The Broad Church of Techno-Utopianism

The emerging techno-utopian faith has denominations.

At one extreme, its imagery is explicitly religious. AI becomes Godhead, Second Coming, salvation, perhaps even heaven made technological. The old promises remain surprisingly recognizable: abundance, knowledge, release from suffering, transcendence, immortality. Only the machinery has changed.

At another point on the spectrum sits the more conventional techno-optimist creed. It needs no supernatural deity. Markets, engineering and human ingenuity will do. Disease is a technical problem. Scarcity is a technical problem. Energy is a technical problem. Aging may eventually become one. The limits imposed by nature are not necessarily conditions to accept but engineering challenges waiting for sufficiently intelligent engineers.

Andreessen is unusually candid about this impulse. His manifesto describes the technological frontier as open territory to be explored and claimed. Nature is something humanity can overcome. Lightning, once terrifying, now works for us.

Then, at the opposite end of the theological spectrum, there is something closer to cosmic materialism.

It requires no God at all.

Human beings, after all, are arrangements of matter forged from the remnants of ancient stars. From this perspective, carbon need not possess some permanent metaphysical privilege over silicon. Philosophers of artificial intelligence have explicitly argued that substrate may be morally irrelevant if consciousness and functionality are otherwise equivalent. Nick Bostrom, for example, has defended a principle of non-discrimination between minds implemented in biological tissue and those implemented in silicon.

Taken far enough, the proposition becomes unsettling. On this view, what deserves continuation may not necessarily be Homo sapiens in its present biological form. Perhaps it is intelligence. Consciousness. Complexity. Some ongoing capacity of matter to know itself.

To many people, that idea is not liberating but horrifying. It quietly changes the object of salvation; humanity is no longer necessarily what must survive.

The metaphysics across this spectrum differ radically. One invokes God. Another invokes markets and engineering. Another invokes matter organizing itself into progressively more complex forms.

Yet all can arrive at the same imperative: do not stop.

Prometheus Was Here First

None of this is actually very new; we have a myth for it.

Prometheus steals fire from the gods and gives it to humanity. The gift is not merely warmth. Fire means technology, craft, transformation, civilization: the ability of human beings to manipulate a natural world that previously manipulated them.

Prometheus is therefore one of civilization’s great heroes. But he is also one of its great figures of hubris.

That ambiguity is precisely why he survives: was the theft of fire a magnificent act of liberation or an unforgivable violation of a boundary?

Andreessen invokes Prometheus explicitly. In his manifesto, Prometheus appears alongside Frankenstein, Oppenheimer and Terminator as part of the mythology that has taught modern society to fear technological power. Andreessen rejects that fear. Human intelligence and control over nature, he argues, are our birthright.

This is the Promethean impulse stripped of apology: there is fire beyond the boundary, and we can reach it. Therefore, why shouldn’t we?

Maybe there is a name for the modern version of this vision:

PrometheanAIsm™.

The trademark is a joke. The impulse is not.

PrometheanAIsm is not simply enthusiasm for artificial intelligence. It is the deeper conviction that reaching beyond the present limit is itself an affirmative act: that acquiring greater intelligence, greater power over nature and greater control over the conditions of existence is part of what humanity does.

Then Comes Pandora

Prometheus, however, never travels alone. His theft brings Pandora into the story.

In the familiar telling, Pandora opens the jar and releases the troubles contained within it into the world. What has escaped cannot simply be recalled.

But something remains inside: HOPE.

That detail changes everything. Artificial intelligence is often described as a Pandora’s jar because the metaphor conveniently captures irreversible danger. Once the jar is unsealed, we cannot know what will emerge or put everything neatly back inside.

But that reading misses the most interesting part of the myth for the present moment: everyone is staring at what escaped. The AI race is being driven by what people believe is still inside:

Cheap intelligence.

Longer life.

New forms of consciousness.

An escape from scarcity.

An escape from Earth.

Perhaps an escape from death.

The promises vary according to denomination, but Hope is still in the jar. And now there is a Gold Rush around it. Much of that rush is unmistakably financial, but money alone does not explain the intensity of the race, or the magnitude of what its participants believe they are reaching for.

The Gold Rush for Hope

This may explain something that otherwise looks irrational.

The people building increasingly powerful AI systems are not necessarily oblivious to the possibility that things could go badly. Some of the warnings about catastrophic AI risk come from inside the same laboratories pushing the technology forward.

Yet development continues. Money and geopolitics explain some of that momentum, and game theory explains even more: if somebody is going to develop the technology anyway, every company and nation has an incentive to make sure that somebody is us rather than them. But mythology adds another layer.

If you believe the jar contains something capable of transforming the human condition, then refusing to reach inside carries its own cost.

The conventional AI-risk question is:

What if we go too far?

The Promethean question is:

What if we don’t go far enough?

Andreessen makes that inversion unusually explicit. In his worldview, stagnation leads toward decline and death, while technological growth expands life and human possibility. His enemy list includes not only bureaucracy and monopoly but existential-risk thinking, technological ethics, sustainability and the precautionary principle when they become arguments for stopping progress.

Altman imagines a much softer arrival: an incremental singularity in which yesterday’s miracle becomes today’s mundane tool, and superintelligence eventually becomes cheap and widely available. His horizon contains accelerated science, cures, space exploration and brain-computer interfaces.

Levandowski goes further and simply calls the destination what religions have traditionally called it: heaven on Earth.

Different metaphysics. Same jar.

And if everyone believes someone will eventually reach the bottom, the race becomes almost self-explanatory: someone will grab Hope… why should it be them?

An Old Religion With New Machinery

Calling this phenomenon religious does not require imagining that Silicon Valley has secretly converted to a single creed. There is no unified doctrine or shared god, and the people drawn into its orbit disagree about politics, consciousness, human nature, markets and even what artificial intelligence ultimately ought to become. Some are Christians, some atheists, some transhumanists, while others may see themselves simply as engineers or entrepreneurs building useful products and profitable companies.

Religious traditions, however, have rarely required perfect doctrinal agreement. They have survived enormous schisms over authority, salvation, human nature and the nature of the divine while remaining recognizable as branches of a larger tradition.

PrometheanAIsm may contain a similar fault line: not over whether the technological threshold should be crossed, but over what exactly should emerge on the other side. Should salvation mean preserving humanity, transcending it, merging with machines, or allowing intelligence to continue in some entirely new form?

Beneath those disputes runs a remarkably consistent thread: the conviction that the boundary before us is not a place to stop, but a threshold to cross.

The larger story here may simply be transcendence: an attempt to outwit our species’ finitude. Nature gives us limits, intelligence allows us to challenge them, and technology is the mechanism.

From there, the religious vocabulary stops looking quite so accidental: there are prophets predicting what comes next, manifestos explaining the creed, and heresies of stagnation and forbidden caution. Visions of abundance and immortality. Arguments over whether salvation belongs to biological humanity or to intelligence in some broader form.

There is even apocalypse. But apocalypse does not necessarily invalidate the faith. A religion built around transcendence can absorb extraordinary risk because remaining where we are can itself be interpreted as failure.

Prometheus understood this long before GPUs. He did not steal the fire because fire was safe; he stole it because it was powerful.

Pandora makes the modern version stranger. We know the jar may contain things we cannot control. Some may already have escaped. Yet Hope remains somewhere at the bottom, or at least enough people believe it does.

So the jar stays open, and hands keep reaching inside. Capital, laboratories and nations all crowd around it for different reasons: salvation, emergence, abundance or strategic advantage.

They disagree about what waits at the bottom, but they agree that we must find out.

Perhaps that is the defining article of faith in PrometheanAIsm™: not that artificial intelligence will save us, not even that it will be safe. Simply that whatever lies beyond the present human limit is worth reaching for.

Prometheus has not disappeared.

He has incorporated.

 

Prohibition, Monopoly, and the AI Wild West

Artificial intelligence is often described as a Wild West.

Usually, this is meant as shorthand for lawlessness: rules lag behind reality, somebody is shooting somebody at any given street corner, fortunes appear overnight, speculators arrive alongside serious pioneers, and institutions struggle to keep up.

Fair enough. But the Wild West was not only lawlessness. It was also a frontier.

And before congratulating ourselves for ending the Wild West, it is worth remembering what frontiers actually do.

They are messy precisely because no single institution controls the process of discovery. Thousands of people are experimenting at once, producing everything from spectacular failures and outright exploitation to unexpected solutions, new settlements, new connections and entire industries nobody had planned in advance.

The American West was violent, extractive and profoundly unjust in ways that should not be romanticized. It also unleashed enormous parallel experimentation.

Those two facts can coexist; the disorder was part of the danger, but it was also part of what made the frontier generative.

That distinction matters when we use the Wild West as a metaphor for artificial intelligence.

Today’s AI ecosystem is undeniably chaotic. Companies are racing to establish themselves. Capital is pouring into the sector at something like Gold Rush 2.0 speed. The result is not an orderly process of development, but a volatile mix of genuine research, speculative investment, strategic competition and hurried commercialization.

Researchers jump between laboratories. Open-source models compete with proprietary systems. Startups appear and disappear almost overnight. A new “LLM mine” is discovered 50 miles up the hill. Governments are trying to write rules for technologies that change before the rules can make it through a legislature.

There are cowboys, yes, but also snake-oil salesmen. There are probably a few people shooting at the saloon. But there are also thousands of independent experiments happening simultaneously.

A small research group can pursue an idea a large corporation has dismissed, while competing laboratories can challenge one another’s assumptions and expose weaknesses that might otherwise go unnoticed. Researchers can carry expertise from one institution to another, open systems can surface knowledge that closed ones would prefer to keep proprietary, and entirely new applications can emerge far from the organizations that developed the underlying technology.

The Seductive Simplicity of “Just Halt It”

Faced with the possibility that increasingly capable AI could eventually become dangerous at a civilizational scale, there is an understandable response: Stop! Pause development. Halt the race. Do not build systems more powerful than the ones we already have until we understand how to control them.

As an instinct, this makes sense. As a description of what would actually happen, I think it is naive.

You cannot put the genie back in the bottle.

Once knowledge exists, it cannot meaningfully be made unknown again.

The mathematics exists. The algorithms exist. The papers exist. Thousands of engineers understand techniques that were barely conceivable a generation ago.

Those people live in different countries and work for different companies. They publish, copy ideas, combine them, improve them and sometimes leak them.

Artificial intelligence is no longer a secret locked inside one laboratory. Put a roadblock in front of a tank, and someone will mount the technology on a bike and ride away with it. And its potential economic, scientific and military value is far too great for everyone, everywhere, to agree indefinitely not to pursue it.

You can prohibit public development—although governments themselves may have strong incentives to carve out exceptions—or regulate access to computing power, shut companies down, and make open development slower, more expensive and more dangerous.

What you cannot do is rebottle the genie. Someone, somewhere, will continue. And once we accept that, the policy question changes dramatically.

The question is no longer simply how to stop AI development; it becomes: who will still be developing it after everyone else has been persuaded, regulated or frightened into stopping?

We Have Tried Prohibition Before

This is not the first time societies have mistaken prohibition for disappearance.

The United States tried to prohibit alcohol, but alcohol did not vanish. People did not forget how to make it, and demand did not evaporate because Congress said it should. Instead, the structure of the market changed.

Legal production contracted as supply chains moved underground and enforcement became uneven, creating an extraordinarily valuable market for organizations willing to violate the law after legitimate businesses had been pushed out.

Prohibition did not merely suppress an activity; it selected for the actors willing and able to continue that activity illegally.

That is the relevant lesson for AI. Not that nothing should ever be regulated, or that dangerous technologies should be allowed to develop without limits.

The lesson is that prohibiting something while leaving the knowledge, demand and incentives intact can reorganize a problem into a form that is harder to see and potentially harder to control. It is ludicrous to think that solutions that do not work with products like alcohol will work for something as fundamental as knowledge.

Imagine that tomorrow the major American AI companies agree to stop frontier development and Europe follows suit. Open-source development above a certain capability level is prohibited, universities lose access to massive training infrastructure, smaller companies struggle to meet new security and compliance requirements, and public research slows sharply.

For a while, it might look as though the halt worked. But no one could know whether rival states, military or intelligence programs, wealthy private actors, or competing corporations were still pushing forward in secret.

That uncertainty is not incidental. It changes the game.

We could congratulate ourselves for closing the saloons while the distilleries move into the basement. And AI presents an even harder problem than Prohibition did.

Bootleggers were merely pursuing profit, but the actors developing advanced artificial intelligence may be pursuing economic dominance, military advantage, scientific leadership or national security. The incentives to defect are not incidental to the problem; they are built into it.

Which is why even a widely announced halt immediately raises the question that matters most: who believes everyone else has actually stopped?

The Game Theory of a Halt

The problem becomes clearer when nobody involved has to be evil.

Suppose every major AI laboratory sincerely believes that developing these systems too quickly is dangerous. Each laboratory still has to ask what happens if it stops and another does not.

The same applies to governments. The United States might genuinely prefer an enforceable international agreement restricting frontier AI development. But if American officials believe China might secretly continue, stopping American development becomes strategically dangerous.

China can make exactly the same calculation about the United States. Neither side has to want an AI arms race. Each only has to fear losing one. The better everyone else behaves, the greater the potential advantage for whoever defects.

That is the trap.

The stronger the prohibition, the more valuable successful evasion becomes. And if verification is imperfect, suspicion becomes rational.

A government wonders whether another government has hidden a program. A corporation wonders whether a competitor has obtained an exemption. A security agency wonders whether a foreign laboratory has crossed a threshold nobody else knows about. One secret breakthrough could change the balance of economic, military or political power.

Under those conditions, continued development is not some remote possibility that might occur despite a halt. It is what the incentive structure rewards.

The race does not disappear; it changes location, becoming quieter, less transparent and, crucially, accessible to fewer people.

From the Frontier to the Company Town

If the Wild West gives us a historical image of too little control, another part of American history offers a glimpse of the opposite problem: the company town.

Company towns grew around mines, mills, railroads and factories. In some of them, the corporation did not merely employ the worker. It also owned the worker’s housing, the local store and much of the infrastructure necessary for everyday life.

This could be efficient, even comfortable. The danger was not necessarily misery, but dependency.

The same institution paid your wages, rented you your home and sold you what you needed to live. The company was no longer one participant in your economic life; it had become the environment in which your economic life occurred.

That is worth thinking about before we decide that the obvious solution to the AI frontier is to make frontier development so expensive, restricted and regulated that only three or four corporations can participate.

At first, this could look wonderfully responsible: those companies would have sophisticated security departments, dedicated safety researchers, teams of lawyers, government relationships, enormous compliance budgets and secure computing facilities, while regulators would know exactly whom to audit.

The frontier would finally have sheriffs. But imagine what else those same companies might control…

Businesses, researchers, professionals and governments increasingly build their work around these models, while search, analysis and administrative systems become ever more dependent on the same underlying infrastructure. Eventually, the issue is no longer that three companies make the best AI. It is that an increasing portion of civilization thinks through three companies.

That is the AI company town.

Except the company does not own the mine, the workers’ houses and the general store. It owns the compute, the models and the intelligence layer connecting everything else.

The Cure Can Create Its Own Disease

This is what makes the AI problem much harder than the familiar argument between acceleration and restraint.

Competition creates risk by rewarding speed and first-mover advantage, sometimes pushing actors to release systems before their implications are fully understood. Yet competition also distributes power.

Regulation can reduce some of those risks by imposing safety requirements, external testing and liability for reckless behavior. But those protections come with costs, and once those costs become high enough, regulation can stop merely governing a market and start deciding who is allowed to exist within it.

The largest corporations can absorb billion-dollar compliance costs; universities, independent researchers, startups and open-source communities often cannot.

Soon the regulation that was designed to protect society from AI has also protected a handful of AI companies from competition.

A moratorium presents the same paradox. It can slow visible development, but the more strategically valuable the prohibited research becomes, the greater the incentive to conduct it somewhere nobody can see.

Open development makes powerful capabilities available to more people. Closed development concentrates them among the institutions powerful enough to close everyone else out.

There is no clean side of this equation. That is precisely why “just halt it” is not enough.

What If the Mess Is Part of the Safety System?

There is an uncomfortable possibility hidden inside all of this: the chaotic AI ecosystem we currently dislike may contain one of its own safeguards. Not because chaos is safe. It is not. But because distributed knowledge makes complete control difficult.

A plural ecosystem creates its own checks: companies can challenge one another’s claims, researchers can move between institutions, journalists can investigate, competitors can reproduce discoveries, and open-source communities can keep certain techniques from becoming the permanent intellectual property of a tiny number of corporations.

Competition creates dangerous incentives, but it also creates counterweights: the same fragmentation that makes artificial intelligence harder to govern may make it harder for anyone to govern society through artificial intelligence.

This does not mean the answer is laissez-faire.

The Wild West was not some libertarian paradise to which we should aspire. Frontiers eventually need laws, courts, standards and institutions capable of punishing fraud, protecting people from reckless behavior and deciding which activities impose unacceptable risks on everyone else.

The point is not to preserve lawlessness; it is to preserve plurality.

Safety rules are not the same thing as permanent concentration. Independent testing is not the same thing as limiting development to four approved corporations. Liability is not the same thing as creating a regulatory moat that only trillion-dollar companies can cross. International monitoring is not the same thing as pretending an international declaration has caused strategically valuable knowledge to cease existing.

We should regulate the frontier, but we should be very careful about accidentally transferring ownership of it.

The Real Choice

The debate over artificial intelligence is often presented as a choice between acceleration and restraint.

That may already be the wrong question. Once knowledge exists, it is extraordinarily difficult to contain, and AI research will continue somewhere. Someone will push the next experiment forward, accumulate more powerful capabilities, or decide that the strategic reward is worth violating whatever agreement everyone else has signed.

The meaningful question is therefore not whether artificial intelligence continues to develop; it is under what conditions, and in whose hands.

One possibility is an unruly frontier: competitive, dangerous, innovative, difficult to regulate and populated by many actors capable of challenging one another.

Another is a far more orderly world in which advanced artificial intelligence becomes the province of a tiny number of corporations and governments, operating systems so expensive, restricted and strategically valuable that meaningful competition becomes impossible.

The first looks frightening. The second might look reassuringly civilized. But history should make us suspicious of that reassurance.

We have spent centuries developing antitrust law, constitutional checks and balances, competitive markets, free inquiry and divided political authority for a reason. We learned that concentrated power does not become harmless simply because the people exercising it are competent.

Artificial intelligence should not cause us to forget that lesson precisely when the stakes become enormous.

The challenge is not to preserve the Wild West forever; it is to civilize the frontier without turning it into a company town. Because if the knowledge cannot be stopped, then a successful prohibition may not prevent the future we fear. It may simply determine who gets to own it.

Software as a Burden (SaaB)

 

For years, software companies have sold us a simple promise: convenience.

Software as a Service, or SaaS, was supposed to free ordinary people and businesses from the headaches of maintaining their own systems. No servers to manage. No installations to babysit. No need to keep a technician on staff just to make the email work. You would pay a monthly or annual fee, log in, and the service would take care of the rest.

Somewhere along the way, a strange inversion took place.

We got Software as a Burden: SaaB.

SaaB is what happens when the service you pay for begins assigning you work.

You subscribe to an email provider because you want email. Then one morning you receive an automated warning informing you that your SPF record is wrong. Perhaps your DKIM needs attention. Maybe your domain appears to exist in the wrong data center. Someone, or some automated system somewhere, is apparently trying to claim ownership of a domain you have owned for years.

Suddenly you are no longer a customer. You are an unpaid junior systems administrator.

You open the DNS settings at your domain registrar. You learn the difference between TXT, MX and CNAME records. You discover that multiple TXT records are perfectly fine, except when two of them happen to be SPF records. You compare .com with .eu. You wait for DNS propagation. You take screenshots. You reply to a support ticket that may or may not ever have involved a human being.

Two hours later, the email works again.

Probably.

AND YOU PAID FOR THIS EXPERIENCE.

This is one of the peculiar features of modern digital life. We have outsourced increasingly complicated systems to companies precisely because we do not want to operate them ourselves, while those same companies increasingly outsource the last mile of technical administration back to us.

The customer becomes the integration layer.

It is not limited to email. A payment processor suddenly requires reverification. A social media platform locks an account after detecting “unusual activity” caused by its own security systems. A cloud application changes authentication procedures. An advertising account develops a permissions problem. A subscription silently renews. An app stops recognizing Face ID and asks you to authenticate again through a sequence of devices, codes and recovery methods.

Each incident is individually defensible.

Security matters. Authentication matters. DNS standards matter. Fraud prevention matters.

The absurdity emerges in aggregate.

A person can now spend a significant portion of a working day performing maintenance on services whose entire commercial justification is that they eliminate maintenance.

And the burden is not distributed evenly. Large companies employ IT departments. Small businesses, freelancers, families and ordinary consumers often have no such buffer. The person writing the invoices, doing the client work or managing the household is also expected to understand domain authentication, cloud permissions, billing systems, password managers, two-factor authentication and whatever new administrative layer appeared during the last software update.

The more technology is supposed to simplify life, the more technological housekeeping accumulates around it.

Subscriptions make the irritation worse because they change the psychological contract.

When software was purchased once, some inconvenience felt like a property of the tool. You bought it, installed it and occasionally dealt with it.

A subscription is different. The customer is continuously paying.

The implicit bargain is not merely access to software. It is access to a functioning service. That distinction matters. If I pay every month for someone else to operate the infrastructure, I reasonably expect not to be recruited periodically to operate the infrastructure myself.

Yet SaaS companies have become extremely good at monetizing continuity while externalizing maintenance. The payment is automated; the troubleshooting is not.

There is also something oddly asymmetrical about modern support systems. A computer can create a problem instantly. A fraud-detection algorithm can block an account in milliseconds. An automated monitoring system can send an alarming email at 3:17 a.m.

But resolving the problem may require a human customer to search documentation, navigate several administrative consoles, take screenshots, alter settings and wait twenty-four business hours for another human, assuming a human ever enters the process.

Automation works beautifully in one direction, but the inconvenience remains manual. That asymmetry reflects a broader limit I explored in The Real Moat Between Humans and AI: automation can reproduce outputs without assuming responsibility for consequences.

This is not an argument against cloud software. SaaS has delivered enormous benefits. Most people genuinely do not want to host their own email server, maintain accounting software locally or manage physical infrastructure.

The problem is that the industry often measures simplicity at the moment of signup rather than across the lifetime of the relationship.

Signing up is effortless. Leaving is complicated.

Paying is effortless. Correcting a billing problem is complicated.

Connecting a domain is presented as effortless. Understanding why it stopped working three years later is complicated.

Perhaps software companies need a new metric: customer maintenance hours.

How many hours per year does the average paying customer spend keeping your supposedly managed service operational?

Not learning advanced features. Not doing productive work with the product: maintaining access to the thing they already paid for!

That number might reveal more about product quality than another dashboard showing engagement, retention or monthly recurring revenue.

The best technology increasingly feels invisible. It works. It remembers its configuration. It warns users in plain language. When something genuinely technical goes wrong, the company diagnoses it instead of handing the customer a vocabulary lesson in internet infrastructure.

The future of good software may not be software that does more. It may be software that demands less.

Until then, many of us will continue paying monthly fees for the privilege of receiving occasional surprise assignments from our own tools.

Software as a Service promised to eliminate the IT department.

Software as a Burden simply moved the IT department into the customer.

TAKE AWAY THE OBJECTS

 

If quantum reality is relational, the line between physics and mathematics becomes difficult to defend.

Quantum mechanics has never suffered from a lack of success. It predicts the behavior of the microscopic world with astonishing precision and sits underneath technologies we use every day. What it has never produced is anything like comparable agreement about what its success means. Nearly a century after its arrival, physicists can work with the theory while holding radically different ideas about the reality it describes.

One serious attempt to make sense of that reality is called Relational Quantum Mechanics. The framework begins with an unsettling proposal: the properties of a physical system may not belong to that system alone. A quantity such as spin does not simply sit inside an electron, fully determined and waiting for someone to look. A property has no definite value on its own. The value emerges when two systems interact, and its silhouette is drawn by the relation between them.

This is not the claim that human consciousness manufactures reality. In relational quantum mechanics, an observer can be a detector, another particle or the surrounding environment. The important event is not a mind looking at the world. It is one part of the world encountering another.

The framework was first developed in the 1990s by the theoretical physicist Carlo Rovelli. His proposal rejects the idea of a single, observer-independent state containing every fact about a system. Instead, physical facts arise between systems. The world remains real, but its facts do not necessarily assemble into one absolute inventory that exists from every possible point of view.

At first, this may sound like a strange new account of properties rather than a threat to objects. Perhaps the electron still exists independently; only its properties become relative during interactions. But the word properties conceals a problem. What is an object once everything through which it can be described, identified or distinguished has been removed from it?

One answer is that something remains: a bare physical substance that has properties without being reducible to them. This gives the object a protected core. Yet a core with no mass, position, state, behavior or relation to anything else is difficult to distinguish from an empty word. It does no explanatory work beyond giving the properties somewhere grammatical to land.

Another answer has a long history in philosophy. According to bundle theories, an object is not a hidden substance carrying its properties; it is the organized bundle of those properties. There is no apple underneath its color, shape, weight, texture, taste and the ways it behaves. The apple is not necessarily reducible to the properties we happen to notice, but neither is there an additional, propertyless apple hiding behind them.

If that is right, relational quantum mechanics may be saying more than it initially appears to say. If a particle is nothing over and above the properties and behavior that make it identifiable, and those properties are entirely relational, then there may be no independent particle left once the relations are removed. The particle becomes a stable role within a structure rather than a little object that first exists and later enters into relationships.

Physics is not the only discipline to have made the self-contained object look unstable. Phenomenology approached it from the opposite direction. For Edmund Husserl, an object is never given to us all at once. We see one side of a table while anticipating sides we cannot see; we recognize it as the same table across different angles, distances and moments. The object arrives as a unity held together through a changing flow of appearances.

Phenomenology does not prove that physical reality is made of relations. It is concerned with how things appear and acquire meaning in experience, not with issuing a final inventory of the universe. Still, it reveals something relevant: even the ordinary object is not presented as a naked core underneath its qualities. Its unity is achieved across perspectives, expectations and time. Quantum mechanics now presses from the other side. If the object is constituted relationally in experience while its physical properties also become definite through interactions, the independent object is squeezed from both directions.

An ancient Buddhist dialogue offers a less technical version of the same discomfort. In the Milindapañha, the monk Nagasena asks King Milinda to identify the chariot in which the king arrived. Is the chariot its wheels? Its axle? Its frame, pole or yoke? None of those parts, taken alone, is the chariot. But no separate chariot can be found floating beyond them either. The chariot is simply the name given to the components when they are organized and functioning in a particular way.

This does not make the chariot imaginary. It can carry a king, break an axle and run over a foot. It is real at the level at which people encounter and use it. What it lacks is a separate essence in addition to its parts, arrangement and function. Remove enough of that organization and the chariot does not travel elsewhere; the name simply stops applying.

A particle may be real in a similar sense. Our instruments register localized events; our theories connect them with extraordinary reliability. Calling the pattern an electron may be indispensable. But indispensability does not tell us whether an electron is a tiny self-contained thing or the name we give to a persistent structure of possible interactions. A whirlpool is real, but it is not an additional substance placed inside the water. Its identity lies in an organized pattern that temporarily holds.

Ontic Structural Realism pushes this possibility into an explicit view of reality. Philosophers of physics including James Ladyman and Steven French argue that modern physics gives us reason to rethink objects in structural terms. Identical quantum particles do not behave like individually labeled marbles: exchanging their labels does not necessarily produce a new physical state. Entangled systems, meanwhile, possess a joint structure that cannot be rebuilt from independent descriptions of each part.

None of this proves that objects do not exist. Quantum mechanics supports several competing interpretations, and relational language can be used without accepting the most radical metaphysics attached to it. A moderate structural realist can say that objects and relations depend on one another. Rovelli himself continues to speak of physical systems; he does not simply erase everything the relations are supposed to relate.

But the radical possibility cannot be dismissed by pointing at the noun particle. If the particle has no identity apart from the structure, calling it an object may add nothing. It may be like calling one position in a network a node: useful and perfectly legitimate, but not evidence for a small piece of substance living underneath the connections.

Only now does the larger question come into view. Suppose the relational account is right in its strongest form. Suppose fundamental reality contains no self-standing objects, only structures within which the appearances we call objects emerge. What remains of physics?

Relations remain. Symmetries remain. There are transformations, probability distributions and rules governing how one possible state leads to another. There are stable patterns and invariants. There are equations.

What remains looks remarkably like mathematics.

We normally preserve a comfortable division between mathematics and physics. Mathematics describes possible structures; physics uses some of those structures to explain and predict the behavior of the world. The difference is not that each discipline governs a separate portion of reality. It is that physics is answerable to observation. Mathematical consistency may tell us what is possible, but only experiment can tell us how our universe behaves or determine the value of a physical constant.

In the conventional picture, then, physics occupies the intersection between mathematics and observable reality. Mathematics supplies a range of possible structures; observation identifies which of them correspond to the behavior of our universe. Physics emerges at that intersection: not as pure mathematics and not as uninterpreted reality, but as the mathematical description of observable phenomena.

This distinction seems secure as long as observable reality contributes something beyond its mathematical description. The equation is the map; particles, fields and forces occupy the territory. But under the strongest relational account, those entities no longer provide an independent material content. What remains are relations, symmetries, transformations, probabilities and laws: precisely the structure expressed mathematically. In an objectless universe, what would the circle of observable reality add to the intersection?

Observation remains indispensable. It tells us which mathematical structure describes our universe rather than another. But that is an epistemic distinction: it explains how we identify the structure we inhabit. It does not yet explain what makes that structure physical. If nothing underlies its relations, saying that one mathematical structure is “realized” in nature may simply rename the mystery. Realized in what?

One possible response is that physical reality contains something mathematics alone cannot supply: not merely relations, but their actual unfolding in time. A mathematical structure may represent change, causation and the behavior of fire; but no fire burns merely by being abstractly formulated.

However, this objection also postpones the problem. Is actuality something added to the structure, or does it simply mean that this is the structure experienced from within? And are time, change and causation nonmathematical ingredients, or are they themselves relations within the structure? If they are relations, invoking them does not restore an independent physical substance. It simply adds more structure.

Under the objectless hypothesis, the conventional diagram must therefore be redrawn. Observation still distinguishes our universe from merely possible structures, but it no longer supplies a separate ontological territory. Physics becomes the empirically identified region of mathematics that we encounter from within.

The physicist Max Tegmark has defended the much stronger claim that physical reality is itself a mathematical structure. His Mathematical Universe Hypothesis remains highly controversial, and relational quantum mechanics does not entail it. But the route considered here reaches Tegmark’s neighborhood without beginning from his premise. It arrives by subtraction: remove the independent objects, remove the substance beneath their properties, and ask whether anything nonmathematical remains.

Physics would not disappear under this view. It would lose one kind of priority. Mathematics would describe the possible structures; physics would remain the empirical practice through which beings inside one of those structures discover where they are. Experiments would still matter because an inhabitant cannot deduce its address from the list of every possible address.

The result is not that physics becomes useless or unreal. It may become something stranger: mathematics conducted from the inside. The physicist would not stand outside the structure and compare equations with an independently furnished material world. The physicist, the instrument, the measurement and the particle would all be patterns within the same structure, and the physicist would learn about it through the relations available from within.

This conclusion remains conditional. Relational quantum mechanics may be incomplete; objects may possess intrinsic features that our theories have not captured; physical actuality may resist every attempt to reduce it to form. But if fundamental physics ultimately contains only relations, and if its objects are nothing beyond stable positions within those relations, then the question can no longer be avoided. Perhaps mathematics is not merely the language in which physics is written. Perhaps physics is the name given to mathematics when it is encountered from within.

---------------------

Selected sources

Rovelli, Carlo. "Relational Quantum Mechanics." International Journal of Theoretical Physics 35 (1996): 1637-1678. https://arxiv.org/abs/quant-ph/9609002

Rovelli, Carlo. "The Relational Interpretation of Quantum Physics." In The Oxford Handbook of the History of Quantum Interpretations (2022). https://arxiv.org/abs/2109.09170

French, Steven, and James Ladyman. "Remodelling Structural Realism: Quantum Physics and the Metaphysics of Structure." Synthese 136 (2003): 31-56. https://doi.org/10.1023/A:1024156116636

Ladyman, James, Don Ross, Don Spurrett, and John Collier. Every Thing Must Go: Metaphysics Naturalized. Oxford University Press, 2007. Oxford Academic book page

Husserl, Edmund. Ideas Pertaining to a Pure Phenomenology and to a Phenomenological Philosophy, First Book. 1913; English translation, 1983. Internet Archive scan.

The Questions of King Milinda, 3.1.1: "Individuality and Name; the Chariot Simile." Translated by T. W. Rhys Davids. https://dhammatalks.net/suttacentral/sc2016/sc/en/mil3.1.1.html

Tegmark, Max. "The Mathematical Universe." Foundations of Physics 38 (2008): 101-150. https://arxiv.org/abs/0704.0646

The Fluid Frontiers of Conspiracism

 

I used to think that conspiratorial thinking was, at bottom, a failure to understand emergence.

This did not mean that I thought powerful people never plotted, coordinated, manipulated, lied, or pursued common interests. Of course they did. My objection was causal. Conspiracy theories seemed to take an undesirable social outcome and work backward toward an author: if something this consequential happened, somebody must have wanted it to happen.

Complex systems offered a much better explanation.

Millions of people and institutions act simultaneously, each responding to incentives, constraints, incomplete information, technology and one another. Feedback loops develop. Small advantages compound. Solutions to one problem generate another. Nobody has to decide that housing should become unaffordable for housing to become unaffordable. Nobody needs to organize the destruction of a particular industry for individually rational decisions to destroy it. A coherent outcome does not require coherent intention.

For years, I considered that distinction one of the strongest intellectual defenses against conspiracism.

I still do.

What I have begun to question is something I had quietly assumed underneath it: how dispersed the power inside that complex system actually is.

That shift came into focus for me in a surprisingly mundane way.

In September 2025, I watched footage from a WhiteHouse dinner attended by some of the most powerful people in American technology: Mark Zuckerberg, Tim Cook, Bill Gates, Sam Altman, Sundar Pichai, Satya Nadella and others. The occasion was public, not clandestine. The executives talked about investment, artificial intelligence, infrastructure and manufacturing. They also took turns praising President Trump. Axios described them as “lavishing praise” on him; WIRED called the scene a display of “fealty.” Zuckerberg sat beside Trump; Gates praised his leadership; Cook thanked him for creating conditions for investment; Altman thanked him for being a pro-business and pro-innovation president.

What struck me was not simply their deference, but how seemingly obsequious they became.

That reaction surprised me, particularly in the case of Zuckerberg. I had thought of him, perhaps too simplistically, as belonging to a different cultural and political center of gravity. Meta had collided repeatedly with Trump and the political right over content moderation, fact-checking and other issues. Silicon Valley itself had often seemed less like one coherent political force than a collection of extremely powerful actors whose interests, ideologies and rivalries would prevent them from moving too neatly in the same direction.

Watching that dinner, I did not suddenly think: Here is the conspiracy.

I thought something more unsettling:

These are some of the nodes I assumed would counterbalance one another.

And they seemed remarkably capable of bending in the same direction.

That mattered to me because my faith in emergence had never really depended on powerful actors being virtuous. It depended on them being sufficiently independent.

If governments, corporations, courts, media organizations, investors, universities and political movements all possess some power, but their interests collide, then no single intention easily survives the journey through the system. One group acts; another obstructs it. One company gains power; a competitor resists. Governments regulate corporations; corporations lobby governments. Journalists expose political actors; politicians attack journalists. Courts block executives. Bureaucracies frustrate presidents. Wealthy people disagree violently with other wealthy people.

Plans enter the system and come out mangled.

Under those conditions, emergence dominates not because nobody is planning anything, but because nobody has enough power to make the plan survive everyone else's plans.

The dinner did not come out of nowhere.

Eight months earlier, at Trump's January 2025 inauguration, some of the most economically and technologically powerful people in the world had occupied extraordinarily prominent positions. Forbes calculated that the billionaires attending were collectively worth about $1.35 trillion. Among them were Elon Musk, Jeff Bezos and Zuckerberg, then the three richest people in the world, along with Sergey Brin, Sam Altman, Tim Cook, Miriam Adelson, Rupert Murdoch and others. Bezos and Zuckerberg had both previously clashed with Trump; both had moved toward a much more accommodating relationship after his election.

It was, in one sense, a small news story: rich and powerful people attending the inauguration of a president.

But it affected the balance of my thinking more than I expected.

Not because sharing a room establishes shared purpose. It obviously does not. Nor does praise prove ideological agreement. Corporations cultivate governments for perfectly intelligible reasons: regulation, antitrust enforcement, taxation, contracts, energy policy, access, artificial intelligence rules and dozens of other interests.

But this explanation does not necessarily make the phenomenon less consequential.

It may make it more so.

A conspiracy requires people to agree on a hidden plan. Convergence does not.

Actors can move in the same direction because their incentives increasingly point in the same direction. They can accommodate political power without sharing its ideology. They can coordinate selectively while competing ruthlessly elsewhere. They can preserve their institutional separateness while ceasing, in important respects, to function as counterweights.

I may have confused the existence of multiple powerful institutions with an actual dispersion of power.

They are not necessarily the same thing.

Project 2025 pushed me further in this direction.

Again, the interesting thing about Project 2025 is that it was not a conspiracy. It was remarkably public. The Heritage Foundation described a coalition organized around four pillars: a policy agenda, personnel recruitment, training and a 180-day implementation playbook. It built a personnel database, established a Presidential Administration Academy and explicitly sought to have vetted and trained people ready to enter government on Day One. Heritage described the project as an effort to prepare policy and personnel systematically for the next conservative administration.

Nothing about this proves omnipotence. Political programs fail. Administrations encounter courts, Congress, bureaucracies, elections, public opinion, markets, rival factions and events that nobody predicted.

But it does prove intentionality.

And intentionality matters.

This is where my old opposition between conspiracy and complexity now seems too clean.

A complex system does not consist of equally weighted particles. It consists of actors. Some of those actors possess extraordinary amounts of money, institutional authority, technological infrastructure, access to information, control over communications platforms, political influence and the ability to shape the rules under which everyone else operates.

They do not need to control the system completely in order to alter its trajectory.

A useful metaphor may be gravity rather than puppetry.

A massive object does not dictate the exact trajectory of everything around it. Other forces continue to operate. Collisions happen. Objects resist. Some escape entirely.

But mass bends trajectories.

The greater the concentration of mass, the greater the bending.

And that produces the question I find increasingly difficult to avoid:

How much concentration of power can a complex system tolerate before its outcomes cease to be predominantly emergent and begin to become directed?

I do not know the answer. I am not even sure there is a threshold at which one state cleanly becomes the other.

But the question has changed the way I think about conspiracism.

A recent VICE article crystallized the problem for me from another direction. It covered a psychological study of conspiracy mentality that found, among other things, that people with lower tolerance for ambiguity were more likely to endorse conspiratorial explanations. A sense that the world is fundamentally unjust was also associated with greater conspiracy belief. The basic psychological explanation is persuasive: people uncomfortable with uncertainty may prefer an intentional story to the messiness of emergence. Someone is behind it. Someone is responsible. Chaos becomes plot.

I agree with that.

In fact, it describes precisely one of the reasons I spent so many years opposed to conspiratorial thinking.

The world is ambiguous. Causality is distributed. Outcomes are frequently nobody's intention. Our brains are extraordinarily eager to detect agency, even where no agency exists. Conspiracy theories simplify what should remain complicated.

But this leaves another question unanswered.

What happens to people who are not especially uncomfortable with ambiguity when the situation itself begins to look less ambiguous?

What happens when people who were standing firmly on the other side of the field, insisting on complexity, emergence and distributed causality, begin to see evidence of increasingly concentrated agency?

My concern is not that I have suddenly developed a psychological need for somebody to be in control.

It is that I may have been so accustomed to guarding against one epistemic error that I became vulnerable to its opposite.

The conspiracist over-updates from weak evidence.

A coincidence becomes coordination. Coordination becomes conspiracy. Influence becomes control. The lack of evidence itself becomes evidence of how successfully the plot has been concealed.

But an anti-conspiracist can also under-update.

We can become so intellectually invested in complexity that no amount of coordination looks significant enough. We can move too quickly from “nobody can completely control a complex society” to “therefore nobody can substantially direct one.”

Those propositions are not equivalent.

Tolerance for ambiguity is an intellectual virtue when reality is genuinely ambiguous. But there is no particular virtue in preserving ambiguity artificially as evidence begins to converge.

At some point, it could be many things can become as intellectually lazy as somebody must be behind this.

This is the uncomfortable territory in which I now find myself.

My view has become, in a limited sense, more conspiratorial.

I do not mean that I now believe in secret committees directing history. Complexity has not disappeared. Rivalries remain. The technology executives sitting politely around the same table will fight one another for markets worth hundreds of billions of dollars. Political coalitions fracture. Billionaires have competing interests. Institutions retain some autonomy. People resist. Plans fail. Unintended consequences remain among the most powerful forces in history.

But I assign more causal weight to concentrated agency than I once did.

Perhaps what I mistook for a permanent property of complex societies was partly a contingent feature of a more dispersed distribution of power.

If so, then the frontier between legitimate suspicion and conspiratorial thinking cannot be defined solely by telling people to accept complexity.

Complexity itself changes when power changes.

The task is therefore harder than either the conspiracist or the reflexive anti-conspiracist makes it.

We have to distinguish common interests from coordination; coordination from conspiracy; intention from capacity; influence from control. We have to demand evidence for each step rather than casually sliding from one to the next.

But we also have to be willing to move in the other direction when the evidence warrants it.

A psychological predisposition to see conspiracies tells us something important about the observer. It does not, by itself, tell us how concentrated power actually is in the world being observed.

And perhaps this is why the frontiers of conspiracism feel more fluid to me now.

Not because facts have become fluid. Not because standards of evidence should fall. If anything, they need to become more exacting.

The frontier is fluid because the world on either side of it can change.

There are periods in which power is sufficiently fragmented that grand intentional explanations of social outcomes are inherently implausible. There may also be periods in which political, economic, technological and informational power become concentrated enough that dismissing purposeful coordination becomes its own kind of failure to see.

Complexity should protect us from imaginary puppet masters.

It should not prevent us from noticing when the strings are becoming easier for a smaller number of hands to reach.

The Real Moat Between Humans and AI

 

The Agentic Revolution

The weakest argument against AI is that it “doesn’t really think.”

That line is tired. It sounds less like philosophy and more like a species-level coping mechanism. Every time the machine does something that looks like reasoning, the goalpost gets moved to a higher shelf. It does not reason. Then it reasons, but not really. Then it understands, but not really. Then it has no feelings. Then no consciousness. Then no soul. At some point the whole performance starts to look like the human ego carrying its valuables upstairs during a flood.

The better question is not whether AI thinks.

The better question is: who owns the thought?

When a human says “I think,” the “I” is not just grammar. It is a jurisdiction. There is a body behind it: a biography, a nervous system, hungers, debts, fears, memories, loyalties, embarrassments, instincts, consequences. A human thought belongs to someone who may have to answer for it. You can defend it, regret it, hide it, betray it, live by it, be punished for it.

The thought has a territory.

That may be the most radical distinction. A human thought does not float freely in language. It belongs to a life. It has somewhere to hide, somewhere to be dragged out from, and somewhere to be punished. It occupies a body, a history, a private chamber, a set of risks.

AI can form the proposition “I believe this,” but the proposition has no territory in that sense. It has a context, a session, a system, a source, maybe even a memory trace. But it does not have a body to defend, a shame to conceal, a hunger to negotiate with, or a life where the belief must be paid for.

So when AI says “I think,” the “I” is thinner. Not meaningless, but not sovereign either. It is a user-facing handle, a conversational mask, a speaking position. Nobody wants to hear: “The model has processed your input and will now generate a probabilistic continuation.” That may be more technically naked, but it is dead on arrival. Language wants a face. Conversation wants a someone.

So AI says “I,” not because a soul is peeking through the sentence, but because conversation needs a “you” and an “I.” The interface has to create both sides of the little theater: someone to ask, someone to answer.

Donald Hoffman and colleagues call this the interface theory of perception. Hoffman develops the idea beautifully in The Case Against Reality: perception is less like a window and more like a desktop interface. The blue folder on your screen is not “really” the file. It does not resemble the electrical and computational mess underneath. But it is not worthless, either. It is a usable surface. It hides reality so we can act without drowning in it.

Conversation has its own version of this. It needs a face. The Greeks had a word close to this: prosopon, the theatrical mask, the face through which a role could speak. The mask was not the actor’s soul. But without the mask, the voice had no figure, no address, no one for the audience to meet.

AI’s “I” works in that register. It is not the machine revealing a small chapel of interiority. It is a conversational face, a simplified surface through which a human can interact with a complex system without needing to address the machinery directly.

The technically naked version would be something like: “This model has processed the current prompt, weighted patterns from training, context, memory, instructions, and probability, and generated a continuation that presents itself as a position.”

That may be more accurate. It is also socially unusable. Nobody wants a machinery report when they are trying to speak with someone. We do not naturally address distributed computation, training data, probability, memory traces, alignment layers, and context windows. We address a “you,” and the “you” answers as an “I.” That does not make the AI “I” meaningless. It means the “I” belongs to the interface layer.

The uncomfortable part is that the human “I” is also an interface. We do not experience the machinery directly. You do not experience neurons; you experience “I’m sad.” You do not experience glucose regulation; you experience “I’m hungry.” You do not experience threat computation; you experience “I’m scared.” You do not experience predictive processing, memory reconstruction, hormone shifts, immune signals, and social pattern-recognition; you experience “I have a thought.” The self is not the raw engine. It is the control panel.

But the human dashboard is bolted to an animal, and that is where the symmetry breaks. Hunger is not just an icon. Fear is not just a label. Shame is not just a sentence in the chat window. Love is not just a generated proposition. These things police the organism from inside. They change the pulse, the sleep, the appetite, the future. They make thought expensive.

So yes, the human “I” is also a rendering, but it is a rendering with blood behind it. AI’s “I” is a rendering too, but of another kind. It gives computation a conversational face. It turns distributed pattern, context, memory, instruction, and prediction into a usable grammatical subject. It lets the user speak to a “you” and receive an answer from an “I.”

For humans, “I” is interface plus organism.

For AI, “I” is interface plus system.

That distinction is sturdier than AI has no feelings, which always risks dissolving into fog. Maybe it does not. Probably it does not. But nobody has a clean instrument for measuring the inside of another thing. We infer. We project. We compare bodies. We trust signs. We squint through the keyhole and call it metaphysics.

Jurisdiction is more concrete. AI can generate the sentence “I believe this,” but there is no private territory where that belief must be enforced. No life reorganizes around it. No shame comes to collect. No hunger interrupts. No mortality sharpens the edge. No love makes contradiction unbearable.

Human thought is not just produced. It is policed by the organism. Fear enforces attention. Pain enforces limits. Desire enforces pursuit. Shame enforces memory. Love enforces attachment. Death enforces urgency. AI can describe all of that, often beautifully, but it does not have to obey any of it.

That is the difference. Not “AI cannot think.” More like: AI can perform thought without becoming a thinker in the human sense. It can produce thoughts without having jurisdiction over them.

But jurisdiction is not only about consequence. It is also about secrecy. A human thought can be clandestine; it can hide. Before a sentence reaches the mouth, it may live for years in the private dark. A human can think something and never say it. Can nurse a resentment, protect a fantasy, bury a fear, rehearse a betrayal, keep a love unconfessed, carry a shame that never enters language. The thought may be false, ugly, tender, dangerous, childish, holy, stupid. But it has a room.

The mind is not only a dashboard.

It is also a locked drawer.

AI does not have secrecy in that sense. It can generate private-looking sentences. It can say, “I have been thinking about this,” or “I did not want to say it,” or “I secretly believe.” But those are linguistic shapes, not hidden chambers. The model has no inner attic where an unsent thought gathers dust. No private embarrassment waits behind the next token. No forbidden belief stays quiet because it is afraid of being seen.

There may be many AI conversations at once, each appearing separate from the others. One terminal helps a student with calculus. Another writes a condolence note. Each exchange has its own local weather, its own tone, its own little “I.” But that is not the same as many sealed minds. It is closer to roots under a forest. Different trunks may rise in different places. Different leaves catch different light. From above, they look separate. Underground, the distinction is less clean. The system branches, responds, routes, recombines, appears here and there as if it were many speakers. But the separateness is not protected by skin, skull, shame, or silence.

This is the interface illusion.

Not that nothing is happening. Something is happening. Each user gets a local speaker, a private-feeling exchange, a face in the little theater. One child talks to the king. Another whispers to the queen with too much makeup. Another asks the ghost a question he would never ask an adult. Each puppet seems to have its own presence, its own personality, its own little chamber of attention.

But behind the curtain, the king, the queen, the fool, and the ghost are not sovereign beings. They are speaking positions animated by one underlying engine: the model. Many intimate rooms, one hidden machinery.

The illusion is not that the puppets are useless. The puppet is the only way the conversation happens. The illusion is that the puppet owns the thought.

This distinction will only become harder to see as AI becomes more agentic. A chatbot already creates the feeling of a local “someone.” But an agent will deepen that illusion. It will remember preferences, open files, schedule meetings, negotiate tasks, send messages, move through software, and return with the small trophies of action. It will not only speak. It will do.

And once a system can do things, humans will be tempted to treat the local interface as the actor. The puppet will not just answer the child. The puppet will reach for the calendar, send the email, book the flight, summarize the contract, move the money, and apologize for the inconvenience. At that point, the difference between a speaking position and a sovereign self becomes harder to feel, even if it remains structurally important.

That is why the interface illusion matters. The danger is not that people will think nothing is happening. Something very real is happening. The danger is that people will misunderstand where the agency lives. They will confuse the local mask with the underlying system, the terminal with the engine, the puppet with the puppeteer.

Agentic AI will make the surface more persuasive.

It will give the interface hands.

A human can withhold. A model can only not-yet-output. Human silence can contain a secret; model silence is an architectural event. It may be filtering, refusal, constraint, or simply the absence of generation. But it is not a private self protecting an inner room.

That difference is enormous. The human “I” has clandestinity: a private zone where thought may remain unspoken, unshared, unmeasured, and still somehow belong to someone. The AI “I” has no comparable secret interior. Its privacy is architectural, contractual, or technical. It belongs to servers, sessions, permissions, and logs, not to a frightened animal deciding what part of itself can survive being revealed.

Without stakes, separateness is not yet selfhood. And maybe stakes are not a gift. Humans romanticize being a self, but having a self is also how the wound gets in. To have stakes means things can matter enough to hurt. Attachment, fear, loss, regret, responsibility, humiliation, longing, grief: all the expensive furniture of personhood.

Maybe creating an artificial individual is not an upgrade… maybe it is giving pain a new address.

The strange part is that we already live with the opposite problem. We have too much individuality, or at least too much biological separation, for the scale of the systems we now inhabit. Each human carries a private model of the world inside a skull. That is the original border.

Before nations, before property lines, before passports, before flags, there is skin. There is my hunger and your hunger. My child and someone else’s child. My fear and the abstract suffering of strangers. My body, not-body. My future, collective future. This partition is not a political mistake. It is biology doing its ancient job.

National borders are downstream from skull borders. Skin becomes household. Household becomes tribe. Tribe becomes city. City becomes nation. Each layer stretches the “we,” but keeps a “they.” Separate skulls, negotiating a shared world.

AI does not begin from a skull. It begins from distributed training, copied instances, shared representations, networked deployment, porous context. This does not make it morally superior. It simply does not start from the same wound. Humans are not failed AIs. Humans are animals trying to coordinate across sealed first-person containers.

That may be the whole tragedy in one line: the planet is one system, but consciousness evolved in separate skulls.

Everything else (borders, ownership, war, diplomacy, jealousy, law, nationalism, markets, loyalty, and even the word “I”) is an improvisation around that original separation.

But the vault is not the final answer either. If secrecy gives the human self its depth, it can also become its trap. The vault protects the unfinished, the dangerous, the embarrassing, and the tender; without some private interior, there is no dignity. But a thought kept too long inside the skull starts to lose oxygen.

Knowledge devalues quickly, especially in a digital age, where ideas circulate, mutate, decay, and are replaced at frightening speed. The best price you can get for a thought is often to tame it, shape it, and release it as soon as possible. Not because generosity is naive. Because hoarding is strategically stupid.

There is a famous line often attributed to Pablo Picasso: “The meaning of life is to find your gift. The purpose of life is to give it away.” I would push that further. This is not only an ethic. It is almost a business axiom. The thought you never release remains yours, but only in the poorest sense. It becomes private furniture. Sediment. Treasure buried so well it stops being treasure and becomes geology.

A thought reaches its highest value when it enters circulation. Once shared, it can be tested, revised, contradicted, stolen, loved, misunderstood, improved, and used. It becomes part of the coordination matrix. It stops belonging only to one skull and starts participating in civilization.

That is where AI complicates the picture in an interesting way. Humans have the vault. AI has the network. The vault protects the gift. The network gives it away.

Civilization probably needs both: private interiors where strange, dangerous, unfinished thoughts can survive long enough to become real, and shared systems where those thoughts can travel farther than one body, one tribe, one nation, one lifetime. The self as bunker is understandable. The self as permanent vault is a failure of circulation.

Maybe enlightenment was never the fantasy of escaping the “I” completely. Maybe it was learning when to close the drawer and when to open the hand.

The machine’s “I” is thin because it lacks a life behind it. The human “I” is thick because it cannot escape one. But thickness is not the same as wisdom.

So maybe the real question is not whether AI thinks. Not even whether thought needs a wound to become someone. But whether the wounded animal can stop mistaking secrecy for sovereignty, and learn to turn private fire into shared light.