Agency, Compounding, and the Argument Over AI Feudalism

A recurring anxiety in contemporary technology discourse asserts that advanced artificial intelligence will irrevocably freeze human hierarchy. The argument, commonly articulated as a warning of technological feudalism, claims that superintelligence will terminate social mobility: because those with the largest capital reserves will purchase the vast majority of compute, raw capital will outcompete human talent and labor, collapsing meritocracy into a permanent oligarchy. A frequent tactical conclusion drawn from this fear is that the present moment represents the last closing window for "ordinary people" to make a class leap by acquiring hard assets before the drawbridge is pulled up.

Yet the premise that a historical door is closing rests on an ontological misunderstanding of human mobility. Class-leap opportunities are never closed by technological decree, because after an individual executes a leap, they are no longer "ordinary." There is no such thing as an opportunity reserved for an abstract demographic category; the individuals who reach the frontier almost invariably climb from near the bottom through relentless, self-directed error correction.

These viewpoints appear as an empirical disagreement over historical timing and technological windows. Yet examining the underlying mechanics reveals that the disagreement is not about compute or timelines. It is about where agency sits, and how the placement of agency governs an individual's rate of compounded improvement.

AI as Recursive Multiplier: The Fluid Spectrum of Agency

Artificial intelligence possesses zero independent agency. It does not desire, choose, initiate, or climb. Through sophisticated mathematical training, it can fluidly emulate goal-directed behavior, planning, and intentionality. Yet this emulation remains strictly initiated, parameterized, and guided by human distinguishing. The machine has no intrinsic will; it operates as a high-bandwidth technological medium that accelerates whatever causal reinforcement loops are brought to it.

Why, then, does the machine appear so menacingly autonomous? Because of an ancient, recurring cognitive inversion: the mass that falls behind systematically re-allocates its own un-owned agency onto the tool itself.

This psychological projection is not unique to artificial intelligence; it has manifested across every major technological shift in human history:

• Ancient populations projected their own moral and physical agency onto stone idols and sacred texts, treating human-crafted artifacts as self-willing deities;
• The Industrial Revolution saw workers project independent malevolence onto mechanical looms, steam engines, and assembly lines, treating mechanical clockwork as a living oppressor;
• Modern societies routinely re-allocate agency onto "the market," "the corporation," or "the bureaucracy," speaking as if these legal fictions possessed independent consciousness and intent.

Whenever a technological paradigm shifts faster than average adaptive habits, those who detach from the underlying causal mechanics externalize causality outward. They treat the tool as an autonomous master, absolving themselves of the responsibility to update their own models. The contemporary panic over "AI feudalism" is simply the newest iteration of this historical habit: confusing the machine's rapid emulation for a sovereign conqueror.

Importantly, the cybernetic posture an individual adopts toward tools is not a rigid dichotomy dividing humanity into two fixed castes. Every conscious Mind is inherently running a sovereign loop, but each of us distributes that agency with varying degrees of externalization across different contexts:

• In one area of life—such as institutional routines, standardized employment, or algorithmically curated feeds—an individual may choose to externalize their decision-making and accept preset defaults;
• In another area—such as a specialized craft, high-conviction investment, creative production, or philosophical inquiry—the same individual may maintain radical, unbuffered sovereignty;
• The two extremes (pure sovereign ownership versus total passive externalization) are theoretical boundary poles; every human being lives somewhere along a continuous, multi-dimensional spectrum.

Crucially, an individual's position on this spectrum is never static, predetermined, or frozen.

At any moment and in any domain, it remains within an individual's sovereign capability to shift their posture along the spectrum—to reclaim their internal loss function, reject externalized excuses, and re-anchor their choices in direct collision with reality.

There is no structural destiny or historical fatalism that can strip this capability away, and no technology—no matter how powerful—that can lock an individual into a subordinate caste without their continuous, active consent. As established across Intelligence belongs only to the Mind and The Generative Topology of Free Variables, AI simply accelerates the specific degree of sovereignty or externalization a person elects to practice in each domain.

Slope Overwhelms Intercept: The Mathematics of Compounding

Over an extended temporal horizon, relative societal standing is not determined by initial position; it is determined by the ranking of compounded improvement rates.

In fundamental terms:

• A starting condition—family wealth, social background, early capital, or geographic luck—is a finite, static starting point;
• An individual's self-directed rate of learning, error correction, and execution is a dynamic compounding rate of improvement;
• Artificial intelligence acts as a high-gain cognitive accelerator.

Over time, any persistent difference in the rate of improvement inevitably overwhelms a finite initial advantage.

In previous technological eras—such as the advent of the steam engine, electrification, or the early personal computer—an individual with a superior improvement rate still faced severe physical and logistical friction. Overtaking well-capitalized incumbents often required decades of physical infrastructure and capital accumulation.

Because artificial intelligence directly accelerates the feedback loop of learning and execution, it multiplies the compounding rate itself. The time required for an active, self-directed individual to surpass a massive initial capital deficit is drastically compressed. Rather than cementing a permanent feudal divide, artificial intelligence actually erases initial starting disparities faster than any technological shift in human history.

A lean, highly focused builder or small outsider team maintaining an unbuffered update loop can now iterate, ship, and outmaneuver bureaucratic corporate monopolies in a matter of months rather than decades.

This dynamic exposes why the tool itself is never the decisive factor: the amplifier multiplies whatever rate of learning is brought to it. The sole independent variable remains the sovereign individual choice to maintain an uncompromised, consequence-bearing reinforcement loop.

Accumulating static assets without this living update rule remains a frozen starting position; it does not generate future capability. As derived in Having More Is Never the Cause and The Generative Mechanics of Value, Money, and Wealth, possession is merely the lagging residue of past decisions—it is never the generative cause of future growth.

The Cognitive Tax of the Closed-World Model

The primary obstacle to maintaining a high improvement slope is the psychological adoption of a closed-world (zero-sum) frame.

When an individual believes that reality is a fixed, zero-sum game where "the window is closing" or "feudal lords own all the compute," their internal optimization objective undergoes a destructive shift:

1. Instead of optimizing the quality of their next real-world action against physical feedback, they optimize their relative standing against visible peers;
2. Attention shifts from objective error correction to monitoring rivals, defending status, and interpreting others' gains as personal losses;
3. The internal feedback signal becomes overwhelmed by social noise;
4. The learning rate drops, and the improvement slope (m) flattens toward zero.

The closed-world frame then generates its own empirical confirmation: because the individual's rate of improvement has stalled, the world genuinely appears impassable and rigid from their subjective vantage point.

As demonstrated in The model never becomes a second edge and Causality All the Way, the reality an observer chooses to score becomes the reality they can reach—not through mystical intention, but because the internal world model dictates the reward function, the reward function dictates the policy, and the policy generates the actual trajectory.

The Irony of Intervention and the Flattened Loop

This dynamic exposes why institutional attempts to artificially equalize outcomes or legislate class mobility so frequently compound the very stagnation they intend to cure.

When an individual's position is adjusted by top-down administrative fiat rather than earned through the recursive cycle of error correction:

• The recipient's attention shifts to learning the administrative rules of the subsidy rather than the underlying physics of competence;
• The individual's cybernetic loss function is disabled;
• The personal improvement slope flattens, converting a temporary disadvantage into permanent institutional dependency.

External resources and technological tools can alter the starting intercept immediately, and if deployed as genuine leverage, can assist in raising the slope. But this occurs exclusively when the recipient does not treat the external aid as the source of their agency.

This generates a profound systemic irony:

• The individuals who can utilize external tools and capital without surrendering their internal loop are those who least depend on external rescue;
• The individuals who most desperately demand external rescue are the most prone to locating their agency inside the external program, guaranteeing their ongoing atrophy.

Standardized institutional procedures that attempt to eliminate inequality by treating human beings as uniform cases inevitably erase the granular variations in risk-taking, curiosity, and local error correction through which autonomous agency operates, as traced in The Misallocation of Cause in Wealth-Transfer Policies and Price Floors and The Welfare State Illusion: Cost Diffusion, Moral Amplification, and the Atrophy of Agency.

The Selective Blindness of the Static Model: Where Breakthroughs Actually Originate

The claim that artificial intelligence will permanently freeze human hierarchy performs an insidious form of selective blindness. It takes a static, lagging snapshot of current capital deployment and misidentifies it as an immutable law of technological destiny.

If the argument that "raw capital buys compute and permanently locks out everyone else" were true, the leadership of modern AI would have been the exclusive, uncontested domain of the trillion-dollar tech conglomerates whose deep pockets and computational infrastructure dwarfed the rest of the world.

Yet the empirical history of the AI revolution demonstrates the exact opposite:

• The foundational breakthroughs that launched the modern frontier did not originate inside the dominant incumbents; they emerged from outsider labs, agile startups, and lean research teams starting from far behind;
• Organizations that had virtually no comparative capital advantage—from early non-profit research groups to lean algorithmic innovators—consistently outmaneuvered incumbent monopolies, shipping paradigm-shifting models before the corporate giants even recognized the terrain had shifted;
• Across every technological epoch, the forefront of innovation is consistently authored by small, highly focused teams operating with sovereign conviction, not by bureaucratic committees managing massive balance sheets.

Proponents of the "AI feudalism" narrative cherry-pick a narrow micro-phenomenon—incumbents purchasing massive GPU clusters today—and shoehorn that single observation into an oversimplified, closed-world model. They confuse the lagging residue of past capital accumulation with the living, real-time distinguishing that authors the future, and then pronounce their own static projection to be "inevitable."

This is the quintessential error of macro-fetishism: looking at the massive, visible monuments of current capital and forgetting that every monument was built by discrete individuals who started with nothing more than an unyielding reinforcement loop.

The Self-Fulfilling Surrender and the Asymmetry of Intellectual Preaching

The tragedy of the "frozen hierarchy" narrative is that the belief in externalized helplessness is strictly self-fulfilling.

The exact moment an individual adopts the premise that agency resides outside themselves—that technological feudal lords control all outcomes, that capital is the sole determinant of success, or that individual striving is futile—they perform the very action that forfeits their agency:

• By locating cause in external structures, they disable their internal error-correction machinery;
• Their improvement slope (m) collapses to zero;
• They manufacture the precise stagnation, helplessness, and obsolescence their theory predicted.

This dynamic becomes particularly stark when examining the class of public intellectuals, academic theorists, and institutional commentators who broadcast these deterministic prophecies.

There is a profound, structural asymmetry between what intellectuals preach and how they actually operate:

• What is expressed publicly is what the commentator wants others to believe—a rhetorical narrative designed to cultivate audience capture, moral authority, or political leverage;
• What guides actual action is the individual's deepest operational belief at the physical instant of choice.

That is why public intellectuals routinely do the exact opposite of what they preach. They lecture the public on systemic determinism, collective victimhood, and the futility of individual competition—yet in their private careers, personal investments, publication strategies, and parenting, they operate with fierce individual agency, calculated risk-taking, and relentless optimization.

They cannot afford to practice what they preach to the masses, because if they actually organized their personal lives around their own doctrines of structural helplessness, they would instantly suffer the very stagnation, decline, and irrelevance that their followers experience.

The Resolution: Where the Loop Remains

The warning of technological feudalism does not describe an inevitable physical law of artificial intelligence; it describes the sociological consequence of human beings surrendering their internal update loops to centralized platforms and administrative buffers. In a world where individuals abdicate their cognitive sovereignty, capital concentration appears all-powerful.

The tactical observation that delay carries an escalating cost remains valid, because AI functions as a high-gain amplifier that causes both functional and dysfunctional feedback loops to compound at unprecedented velocity.

Yet the foundational invariant remains: opportunity does not belong to a static sociological collective labeled "ordinary people." Because every individual retains the continuous, moment-by-moment freedom to shift from passive externalization to active sovereignty, there is no predetermined ceiling or permanent aristocracy. The moment an individual takes undivided ownership of their loss function, seeks unbuffered feedback, and uses every available tool as leverage to upgrade their predictive model, they cease to be ordinary.

The future of any individual is not determined by the initial coordinate they occupy or the compute reserves of distant corporations. As shown in the extendable horizon, positive-sum agency can be activated from any coordinate in the public register, where daily sovereign choices compound a normal distribution of micro-actions into a power law of realized capability. It is determined by the fluid, sovereign choices made at the moving edge of their own reinforcement loop. AI will accelerate whatever posture is brought to it: a closed, rivalrous posture will compound stagnation; an externalized posture will compound dependency; but a self-directed, consequence-bearing posture will compound generative capability into an unassailable trajectory.