AI Mania: From Tulips to Tokens

TEXT START: Michael Pollan has written about everything from mushrooms to apples and corn, but his coevolutionary story of how tulips and humans shaped each other may have been the most prescient given recent market parallels.

The Dissection

This is an anti-hype essay disguised as an ecological meditation. It treats AI as a self-replicating system, punctures singularity claims with failed autonomy predictions, exposes the material costs of compute, and proposes local or public ownership as a corrective.

It also accidentally documents the Discontinuity Thesis: CEOs attribute layoffs to AI, workers fear the next paycheck, and Claude Code completes a task without remaining “in the loop.” The tool is already beginning to set the pace and define the work. The essay sees the symptom, then retreats into stewardship, transparency, and community ownership.

The Core Fallacy

The text treats consciousness, market mania, and infrastructure limits as if they determine AI’s economic threat. They do not. An unconscious system that performs cognitive work more cheaply is sufficient for displacement. A speculative bubble can burst while the underlying automation survives.

The claim that “relationships and understanding drive value” is an unsupported human moat. If those capacities can be modeled, delegated, or embedded in workflows, they cease to protect mass employment. “Tools amplify expertise” becomes “tools replace expertise” when the tool can execute the objective itself—the Claude Code anecdote proves the point.

The decisive question is not who is using whom in a biological sense. It is who owns the AI capital, who captures the returns, and what happens when most people lose access to economically necessary labor. The article largely avoids that question.

Hidden Assumptions

• Human judgment, relationships, and accountability will remain uniquely human and economically scarce.
• Current model failures and stale knowledge represent durable limits rather than temporary lag.
• Grid constraints, water use, legislation, and local resistance can preserve human-only economic domains at scale.
• Public or community ownership of compute can restore agency rather than merely distribute the output of a labor-displacing system.
• Open weights, local models, and plain-language interfaces will broaden productive participation instead of making substitution cheaper.
• The historical survival of programmers disproves the possibility of eventual cognitive labor collapse.
• Better governance can neutralize competitive pressure to automate.

Social Function

Classification: partial truth serving transition management, with an ideological anesthetic effect.

The essay is right about opacity, extraction, energy, land, water, speculative excess, and the fiction of a permanent human-in-the-loop. It is not simple copium. But it converts productive participation collapse into a more comfortable problem of consciousness, ecological ethics, and ownership design.

That framing suggests AI can be made benign if it is properly sited and publicly controlled. Under DT logic, ownership may determine who receives the surplus, but it does not restore the mass employment-to-consumption circuit. It manages the carcass; it does not resurrect the animal.

The Verdict

This is an intelligent ecological indictment of AI infrastructure and financial mania, but it is not a complete obsolescence analysis. It mistakes the fever—tokens, hype, failed forecasts, consciousness claims—and the smoke—water, energy, opacity—for the engine: competitive cognitive automation.

Frontier labs can fail. Markets can crash. Regulation can delay deployment. Compute can become a public utility. None of that necessarily stops P1, P2, or P3. The article’s final insight is correct: humans become tools of tools when machines set the pace, allocate the work, and capture the returns. The text recognizes that process beginning, then refuses to call it terminal.