The world can't afford to get AI wrong - Interest.co.nz

TEXT START: AI is here, and already having a profound impact on the global economy.

The Dissection

This is an institutional containment memo disguised as transition analysis. It accurately identifies the infrastructure boom, leverage, circular financing, cybersecurity exposure, concentration, inequality, tax-base erosion, and the possibility of substantial job displacement. Then it routes every consequence back into familiar tools: regulation, reskilling, labor-market flexibility, infrastructure spending, safety nets, and international coordination.

The article’s central operation is to acknowledge the machinery of disruption while preserving faith in the continuity of the system being disrupted. “Manage the transition” is the governing euphemism. It assumes the transition has a destination inside capitalism rather than a rupture with its postwar operating logic.

The Core Fallacy

The article treats labor displacement as a reallocation problem rather than a productive-participation collapse.

Its claim that workers may move into new roles assumes that AI will create enough economically necessary, remunerated work to replace what it destroys. Under the Discontinuity Thesis, that assumption fails. Once AI delivers durable cost and performance superiority across cognitive work, firms are compelled to reduce human labor wherever competition allows it. Institutions cannot preserve large human-only economic domains indefinitely while rewarding automation elsewhere.

The result is not merely occupational churn. It is the severing of the mass employment → wage → consumption circuit.

New occupations such as prompt engineering and AI-agent development are implementation niches, not evidence of a scalable replacement labor market. They are also vulnerable to the same automation they help deploy. Reskilling workers for a labor market whose frontier is moving faster than human adaptation is not a bridge; it is training people to chase a retreating shoreline.

The article also confuses productivity with broad prosperity. A positive supply shock can enrich AI owners while reducing labor income, bargaining power, and the tax base. Transfers may preserve consumption, but they do not restore productive participation or give displaced people control over the capital replacing them. That is stabilization of demand, not survival of the old social contract.

Hidden Assumptions

• New jobs will scale roughly with the jobs AI eliminates.
• Productivity gains will diffuse to labor rather than remain concentrated among AI owners and infrastructure controllers.
• Reskilling can outrun accelerating model capability.
• Firms will voluntarily share automation gains instead of using them to compress payroll and increase margins.
• Governments can tax concentrated AI capital before capital mobility, lobbying, and fiscal dependence hollow out that capacity.
• International coordination can restrain adoption despite competitive pressure to automate.
• Financial stability and social stability are substantially the same problem.
• The absence of visible aggregate employment damage so far indicates manageability rather than a lag before displacement becomes legible.
• “Broader adoption” benefits societies broadly, rather than broadening dependency on a narrow ownership class.
• Flexible labor markets are a neutral adjustment mechanism, rather than an efficient machine for liquidating human bargaining power.

Social Function

Primary classification: transition management.

Secondary classifications: ideological anesthetic, elite self-exoneration, and partial truth.

This is not empty copium. Its warnings about financial fragility, cyberattacks, energy demand, concentration, and inequality are real. Its failure is more consequential: it treats those dangers as governable side effects while refusing to follow the labor-displacement mechanism to its structural endpoint.

The policy agenda reassures institutions that they can remain administrators of the future. It converts a possible regime change into a checklist of reforms and converts ownership conflict into a skills deficit. The closing appeal to embrace the challenge together is ceremonial consensus layered over an unequal transition.

The Verdict

This is a competent pre-collapse memo that mistakes damage control for preservation. It sees the leverage, circular capital, labor exposure, and tax erosion, but stops before the decisive implication: AI can destroy the economic necessity of most human labor faster than institutions can invent a replacement for it.

Its prescriptions are lag defenses and hospice care. Regulation may slow financial contagion. Safety nets may preserve consumption. Infrastructure and reskilling may create temporary niches. None of them restore mass productive participation or distribute control of the replacement capital.

Under the Discontinuity Thesis, the article imagines a more productive capitalism with AI. The harder trajectory is a system divided between Sovereigns who own or control AI capital, Servitors who remain indispensable for physical, political, or maintenance functions, and a managed majority sustained through transfers and declining economic relevance. That is not a smooth transition. It is the postwar order dying while its administrators call the symptoms “policy challenges.”