The AI Jobs Debate Is Asking the Wrong Question - Observer Research Foundation

TEXT START: AI has spread through the economy without denting employment.

THE DISSECTION

This is a transition-management memo disguised as a systemic diagnosis. It treats stable aggregate employment as evidence that AI remains mainly assistive, then relocates the danger into adjustment speed and distribution of productivity gains. Its remedies—reskilling, wage insurance, youth guarantees, and competition policy—assume the labour market will eventually generate a new equilibrium. The article acknowledges junior-worker exposure and AI rent capture, but treats them as manageable distortions rather than evidence of a collapsing participation system.

THE CORE FALLACY

It mistakes a lagging aggregate for proof that the employment system survives. “No dent” in employment may describe the pre-collapse phase, while AI is still wide but shallow. Under the Discontinuity Thesis, P1 makes cognitive automation increasingly self-sufficient; P2 prevents stable human-only economic domains at scale; and P3 eliminates economically necessary labour for the majority.

The article presents substitution as merely one branch alongside complementarity. DT treats substitution as the terminal direction once capability, adoption, and competitive pressure mature. Demand elasticity cannot save mass employment indefinitely: cheaper output may expand markets, but it does not guarantee expansion faster than AI eliminates human labour. Training also fails when the technology lowers the cost of performing the trained skill itself. Policy can redistribute income, delay adoption, and cushion consumption. It cannot recreate productive necessity for displaced workers.

HIDDEN ASSUMPTIONS

• Current shallow deployment is treated as a durable condition rather than a temporary adoption stage.
• New demand and new occupations will absorb displaced workers at sufficient scale.
• Human labour will remain indispensable somewhere in the production chain.
• Skills programmes can outrun falling automation costs.
• Competition policy can prevent durable concentration at the model, cloud, and compute layers.
• Stable employment is treated as equivalent to stable productive participation.
• Governments can coordinate quickly enough to manage a transformation the article itself describes as historically rapid.
• Retaining junior roles can be made economically rational even when AI performs their training tasks more cheaply.
• “Eventually” finding a new equilibrium is treated as a historical law rather than an unsupported extrapolation.

SOCIAL FUNCTION

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

The article accurately identifies short-run facts within its own account: aggregate employment has not yet collapsed, AI use is often incomplete, adjustment lags matter, and supplier concentration can capture the gains. Its social function is to convert a possible rupture into an administrative checklist. That reassures institutions that the problem is policy execution, not the destruction of the wage-to-consumption circuit. Transfers, retraining, and competition rules may manage the carcass; they do not restore mass productive participation.

THE VERDICT

The article is competent at describing the waiting room and structurally wrong about the disease. It offers a valid near-term labour-market snapshot, but its long-term conclusion depends on the very assumptions AI is eroding: elastic demand, durable human necessity, and automatic re-equilibration. Under DT logic, this is not a refutation of obsolescence. It is an early-stage report from a system that has not yet admitted what its own mechanics imply.