A cold shower for the AI mania - The Korea Times

TEXT START: By Raghuram G. Rajan

The Dissection

Rajan performs the role of the credentialed establishment skeptic—a former central banker applying conventional supply-chain risk analysis to the AI transition. The piece reads as responsible temperance. It is, in fact, institutional hedging dressed as analysis. He maps the AI supply chain, identifies potential choke points (hardware depreciation, model plateau, political regulation, compute shortage), and concludes that rollout timing and profit distribution are uncertain. A reasonable, measured piece. Utterly insufficient.

The Core Fallacy

Rajan treats this as a rollout logistics problem with a political risk overlay. His frame: how quickly and unevenly will AI deploy, and who captures the margins? He never asks the relevant question: what happens to the mass employment circuit when AI capabilities saturate cognitive work regardless of rollout speed?

He acknowledges AI will displace jobs ("the most important trigger for political intervention would be massive AI-related job losses") but treats this as a political headwind that might slow adoption. The DT inverts this: job displacement is not a slowdown mechanism, it is the terminal mechanism. The political hesitation Rajan identifies is not a bug that delays the collapse—it is the collapse manifesting. Firms won't shed employees outside a recession. So the recession will come, or the displacement will be slower but structurally complete. Either way, the circuit severs.

His escape hatch—"a more limited, careful AI rollout could give firms more time to find labor-augmenting (as opposed to labor-displacing) uses"—is the foundational wish-casting of every DT-skeptic. The assumption is that there exists a stable, scalable domain of human-competitive cognitive work that AI will not saturate. This assumption is not defended. It is assumed.

Hidden Assumptions

1. Augmentation equilibrium exists. He treats labor-augmentation as a natural resting state, a capability humans can maintain. The DT argues this is a temporary, shrinking domain.

2. Political resistance is the binding constraint. Rajan treats regulation and moratoriums as exogenous shocks that might interrupt the rollout. Under DT logic, political resistance is the system responding to the death of its own employment architecture. It cannot save the circuit; it can only manage the corpse.

3. Profits and system survival are the same question. His entire profit-analysis frame (who captures margins: hardware makers, hyperscalers, AI labs) treats the viability of the AI sector as the primary question. The DT treats the viability of the post-war consumption architecture as the primary question. Rajan does not engage this second question at all.

4. Corporate end-users are the decisive actors. He spends significant energy analyzing why corporations haven't fully deployed AI (data organization, brand security, hallucinations). This is a lag-indicator analysis. The reason firms haven't deployed end-to-end is not structural hesitation—it is that the transition hasn't been forced. When competitive pressure saturates (it will), hesitation collapses.

5. "AI advances will likely pay off eventually." The implicit beneficiary in this sentence is never specified. If "paying off" means AI companies generate returns for investors, this may be true. If it means the post-WWII economic order continues to function with mass employment as its primary distribution mechanism, the DT says no.

Social Function

Transition management and expectation calibration. Rajan's institutional position demands he perform "serious concern" without threatening the paradigm. The piece tells establishment audiences: don't worry, this is a rollout problem, not a structural problem, and the smart money will adapt. It reassures the productive class that there is still time to find a place. It reassures AI companies that political resistance is the real risk, not structural obsolescence of their market.

This is prestige signaling wrapped in supply-chain analysis. Rajan's credentials are doing the work: former RBI governor, IMF chief economist, Chicago Booth. The content reads as authoritative precisely because it confirms what the establishment wants to hear—that the AI transition is a timing and margin question, not a civilization question.

The Verdict

Rajan has written a competent, measured analysis that entirely misses the point. He identifies the supply-side risks of AI deployment, treats job displacement as a political problem that might slow rollout, and offers "careful rollout" as the solution. None of this engages the DT's core theorem: when AI systematically replaces human cognitive labor at scale, the mass employment → wage → consumption circuit dies, regardless of rollout speed or profit distribution.

The piece is useful for those who need reassurance that the transition is manageable. It is useless for those who need to understand the structural mechanics of what is actually happening.

Classification: Elite expectation calibration / Prestige-backed reassurance theater.