AI should complement humans, not imitate them | The Asset

TEXT START: To unlock AI’s full economic potential, we must steer it towards expansion.

The Dissection

This is a rescue brief for labor-centered capitalism disguised as historical economics. It concedes that AI lowers the cost of cognition, then assumes that cheaper cognition will generate enough new human work to absorb everyone displaced by it.

The article conflates three different things: more output, more markets, and more human employment. Earlier machines often complemented workers because they automated narrow physical tasks, adoption was slower, and genuinely new human tasks remained scarce. AI attacks cognition itself across many sectors at once. New services may be supplied primarily by AI, not by the workers it displaces.

The text also shifts the problem from ownership to design. If AI produces mass unemployment, it implies that policymakers chose the wrong incentives. That lets capital owners and institutions imagine the catastrophe is a steering failure rather than the predictable result of competitive returns.

The Core Fallacy

The central error is treating expanded demand as expanded demand for human labor. Lower prices can create more consumption while labor’s share collapses. A market can become vastly larger and require fewer people to operate it.

The ATM analogy is structurally weak. ATMs automated a limited function while banks still needed human tellers for a broad range of branch activities. AI is not merely a cheaper tool beside cognition; under the Discontinuity Thesis, it becomes a scalable competitor to the cognitive worker. The historical pattern is therefore not a law. It is a lag-dependent exception.

The article’s second error is assuming that judgment, taste, accountability, and relationships are permanently human economic scarcities. Some may retain value, but they can be simulated, standardized, automated, or reduced to legal and reputational wrappers around machine output. Even where they remain valuable, they may support a small premium class rather than mass employment.

Its third error is assuming that policy can enforce complementarity under competition. A firm that refuses substitution loses cost, speed, and scale to a rival that adopts it. National rules can delay or redistribute the shock. They cannot preserve stable human-only economic domains globally.

Hidden Assumptions

• New human tasks will appear in sufficient volume to absorb displaced workers.
• AI will not automate the supposedly complementary tasks created around it.
• Human judgment, taste, trust, and accountability will remain superior in market-relevant ways.
• Productivity gains will flow into wages rather than capital rents.
• Owners will accept lower returns in exchange for augmentation.
• Governments can coordinate across jurisdictions and enforce those incentives.
• Historical employment absorption will repeat despite simultaneous automation of cognitive work.
• More prosperity will be widely shared without changing ownership or control of AI capital.

These assumptions are asserted, not demonstrated.

Social Function

Primary classification: ideological anesthetic and transition management, with a real partial truth.

The article correctly identifies that automation can create new tasks and that distributional incentives matter. But it uses those truths to make structural collapse sound optional. It offers policymakers a humane vocabulary—augmentation, expansion, capability—while leaving the ownership regime largely untouched. Its message is comforting to elites because it suggests that the wage system can be saved through better steering, even as the competitive mechanism rewards substitution.

The Verdict

This is not a rebuttal of the Discontinuity Thesis. It accepts P1—the falling cost of cognition—then assumes away P2 and P3.

AI can complement selected humans while substituting for the majority at the system level. More markets do not mean more human necessity. Under the DT framework, the likely outcome is a bifurcation: Sovereigns own and direct AI capital; Servitors remain valuable in narrow bottlenecks such as accountability, trust, physical operations, maintenance, logistics, and transition management; the majority lose access to economically necessary labor.

Transfers may preserve consumption. They do not restore productive participation. “Complement humans” is a legitimate policy preference, but without a mechanism that changes ownership and control, it is hospice language for a dying wage system.