John Whitehead | AI authoritarianism: A government without constitutional brakes

TEXT START: “United States AI companies must be free to innovate without cumbersome regulation.”

The Dissection

This is a constitutional alarm built around one valid distinction: fewer government employees can coexist with vastly greater government power. The text argues that automation converts a manpower-limited bureaucracy into a software-scaled surveillance and enforcement system, where decisions become faster, cheaper, less visible and harder to contest.

Its strongest insight is that “bureaucracy without bureaucrats” can be more powerful than ordinary bureaucracy. Its weakness is that it treats the disappearance of human personnel as the central catastrophe while ignoring the economic machinery forcing automation forward. It never seriously addresses who owns the AI capital, who controls the compute, data, energy and logistics, or what happens when the wider population loses productive necessity.

The article sees the new control system. It does not see the economic order that makes the system inevitable.

The Core Fallacy

The text romanticizes human bureaucrats as constitutional brakes. Human beings can hesitate, object and refuse, but they can also obey, abuse authority, rubber-stamp decisions and conceal misconduct. A named bureaucrat is not automatically accountable, and human judgment is not inherently a liberty-preserving mechanism.

Under the Discontinuity Thesis, the decisive mechanism is not simply that officials surrender judgment to machines. It is that AI becomes durably superior across cognitive work, making automation competitively mandatory. Institutions that preserve human-only administrative domains lose speed, cost efficiency and strategic capacity. That is P1 and P2. The resulting collapse of economically necessary human labor is P3.

The article correctly identifies automation as a force multiplier for state power, but misidentifies the cause as moral surrender to technology. The deeper mechanism is ownership and control. AI does not need to become a sovereign actor. Sovereigns use it to make coercion scalable, while responsibility is dispersed across contractors, models, procedures and supposedly objective outputs.

Human review is therefore a lag defense, not a reversal. It can slow, expose or occasionally stop abuse. It cannot restore the mass employment-to-wage-to-consumption circuit or prevent AI capital from displacing human productive participation.

Hidden Assumptions

The text smuggles in several assumptions:

• Human officials are generally independent enough to resist unlawful orders.
• Human involvement creates meaningful accountability rather than ceremonial approval.
• Constitutional procedures can scale at the speed and volume of automated administration.
• Courts, legislatures and regulators will retain effective control over privately owned AI infrastructure.
• AI deployment is mainly a choice of ethics rather than a competitive requirement.
• Preserving human judgment would preserve the old constitutional balance.
• The principal danger is machine opacity, rather than concentrated ownership of the machine.
• Government can retain the benefits of AI while refusing the structural consequences of AI dominance.

The article also treats “humanity” as though it were a stable institutional property. It is not. Human institutions are already hierarchical, coercive and vulnerable to capture. Removing personnel may eliminate some opportunities for resistance, but retaining them does not solve the underlying power problem.

Social Function

Classification: partial truth, transition management and constitutional lullaby.

This is not pure copium. Its warning about machine-speed surveillance, automated eligibility decisions, private contractors and autonomous targeting is structurally credible within the text’s own logic. It identifies a real expansion of state capacity and a real accountability vacuum.

But it channels that danger into nostalgia for human brakes and procedural restraint. It encourages readers to defend the old administrative surface without confronting the ownership structure beneath it. The result is an ideological anesthetic: fear is directed toward black-box software while the decisive struggle over AI capital, control and displaced labor remains mostly untouched.

The article wants constitutional limits to survive the transformation. It does not explain how they survive the concentration of productive power in the hands of those who own and command the automated system.

The Verdict

The article correctly sees that a smaller payroll can produce a larger and more invasive state. It correctly warns that software can multiply surveillance, classification, enforcement and military error beyond human scale.

But it mistakes the removed human intermediaries for the main cause of the crisis. Under DT logic, the deeper rupture is capitalism without workers: AI severs the mass employment, wage and consumption circuit, while the state automates itself to manage the resulting population and preserve concentrated control.

Constitutional brakes, human review and legal obstacles are useful lag defenses. They may slow the machine. They do not reverse the trajectory. The text sees bureaucracy without bureaucrats. It does not see the economic system that is becoming society without productive participants.