Bernie Sanders has AI policy all wrong - Disconnect Blog

URL SCAN: Bernie Sanders has AI policy all wrong - Disconnect Blog
FIRST LINE: In December 2025, Senator Bernie Sanders became the first member of Congress to demand a moratorium on the construction of new AI data centers.

The Dissection

The article is trying to sever AI policy from the industry’s grand mythology. It correctly foregrounds data-center externalities, present-day harms, and the use of AI rhetoric as cover for layoffs and cost cutting. Its political objective is to preserve Sanders’s moratorium while discrediting his belief that AI will fundamentally restructure labor and wealth.

But the article replaces one simplification with another. It treats generative AI as either a civilization-transforming intelligence or fraudulent “autocomplete.” That is a false binary. The relevant question is not whether a chatbot thinks, understands, or possesses consciousness. It is whether increasingly cheap systems can produce economically usable outputs and reduce the amount of human labor required to produce them.

The Core Fallacy

The article commits an ontological-to-economic category error. “It is only pattern prediction” does not establish “it cannot replace workers.” Markets pay for results, not inner comprehension. A system can hallucinate, require supervision, and lack consciousness while still automating enough bounded cognitive work to erode labor demand.

Weak current returns and employer misuse are not proof of permanent failure. They may reflect deployment lag, workflow integration costs, immature infrastructure, or firms using early automation primarily as a cheaper substitute for headcount. The Discontinuity Thesis does not require sentient AI or flawless general intelligence. It requires durable cost and performance superiority across enough cognitive tasks to sever the mass employment-to-consumption circuit.

The article also mistakes visible infrastructure damage for the whole mechanism. Data-center pollution, electricity costs, and water use are real constraints, but a moratorium can delay or relocate capacity without restoring productive human participation. Capital, computation, and automation can migrate. The carcass does not revive because one feeding ground is fenced off.

Hidden Assumptions

• If AI lacks human-like understanding, it cannot create economically decisive substitution.
• Hallucinations are an intrinsic limitation that will remain fatal rather than a constraint managed through verification, retrieval, tooling, or task selection.
• Present productivity disappointment is a ceiling, not a diffusion-stage lag.
• Layoffs attributed to AI are mostly fraudulent excuses rather than an early form of labor substitution.
• Automation must eliminate entire occupations immediately to threaten the employment-wage-consumption circuit.
• Regulation of domestic data centers can materially halt the underlying competitive race.
• Harms such as addiction, deepfakes, cognitive decline, and information pollution are separate from the economic transition rather than its social side effects.
• Listening to victims and skeptical experts is sufficient to understand the system-level trajectory.
• The current form of chatbot deployment is representative of the future capability and cost curve.

Social Function

Classification: partial truth, ideological anesthetic, and transition management.

The article performs a useful demystification of industry hype and identifies genuine material harms. Its anesthetic function is subtler: by portraying AI as mostly fraudulent autocomplete and layoffs as managerial theater, it reassures readers that the deeper threat is exaggerated. It permits opposition to data centers and abusive deployments without confronting ownership of AI capital, competitive displacement, or the loss of economically necessary work.

It is therefore a mirror image of the hype it attacks. Industry propaganda says AI will transform everything through intelligence. This article says AI cannot transform the economy because it lacks intelligence. Both grant the wrong variable decisive status. The decisive variable is substitution at scale under competitive pressure.

The Verdict

The article is right about the externalities and wrong about the terminal mechanism. It exposes the smoke around AI while misidentifying the fire. Non-sentient, unreliable, overhyped systems can still become powerful labor-displacement machinery once cost, integration, and verification improve. A data-center moratorium may buy time or constrain local damage; it does not defeat the Discontinuity Thesis. The article’s skepticism is a partial truth converted into denial of the structural break.