AI Can't Replace Human Judgment - City Journal

TEXT START: As the political salience of artificial intelligence keeps rising, a new book brings much-needed evidence and rigor to an increasingly polarized debate.

The Dissection

The text converts a labor-demand shock into an occupational-reranking story. It treats accountability, trust, judgment, political coordination, and physical presence as durable human moats rather than temporary costs of separation. It makes one useful concession—that lower tiers within professions may be eliminated—then buries its implications under aggregate demand growth, the Jevons effect, and the revaluation of judgment.

The article’s own example gives away the mechanism: a compliance operation requiring 500 workers may require 30. That is not protection. It is labor liquidation with a respectable vocabulary.

The Core Fallacy

The article confuses a surviving task or function with a surviving mass labor market.

Judgment may remain economically valuable without remaining widely employable. AI does not need to eliminate judgment, trust, or accountability; it only needs to let a smaller number of humans supervise vastly more output. The human signature, client relationship, or legal exposure can persist while the surrounding workforce disappears.

“Messiness” is therefore not an AI-proof category. It is a bundle of separation costs. As systems improve, institutions adapt, liability is redistributed, and oversight tools become more capable, those costs fall. The autonomy threshold is not a permanent boundary. It is a moving target, and competition keeps pushing it outward.

The Jevons effect is also misapplied. Cheaper services can expand total consumption while requiring fewer workers to produce each unit. More output does not mean more broad-based employment, wages, or productive participation. Under P1 and P2, expanded demand is captured by AI owners and the humans who control scarce residual judgment. P3 follows: the majority lose access to economically necessary labor even if society consumes more services than ever.

Hidden Assumptions

• New demand will translate into proportional human employment rather than concentrated AI-owned production.
• Legal accountability must remain attached to large numbers of human professionals instead of being centralized in firms, insurers, protocols, or a thin supervisory layer.
• Trust and judgment require the same headcount as before, rather than one trusted human directing machine-produced work at massive scale.
• Institutions will preserve current occupational boundaries instead of redesigning them around AI capability.
• Human presence is a permanent necessity rather than a lag defense imposed by current technology, regulation, and customer expectations.
• Professional credentials correlate closely enough with real judgment to support broad employment.
• A more productive economy will distribute gains widely rather than intensify sovereign ownership and servitor dependence.
• Occupational survival means workers survive, when the relevant variable is the share of workers retained.

Social Function

Classification: partial truth functioning as transition management and ideological anesthetic.

The article is more sophisticated than simple AI denial. It correctly identifies that some judgment, trust, and physical presence will remain valuable, and it correctly warns that survival is not distribution. But it still reassures the professional class that “messy” work is safe and recasts mass displacement as a healthy revaluation of talent. It prepares readers to accept extreme concentration without naming the ownership structure producing it.

The Verdict

The article sees the corpse’s remaining reflexes and calls them a viable organism. Its microeconomic observations are credible; its systemic conclusion is not. Judgment, taste, trust, and accountability may survive as functions while the number of humans paid to perform them collapses. The endpoint is not automatically a healthier or more egalitarian economy. It is a narrower sovereign class, a smaller indispensable servitor class, and a displaced majority maintained as consumers through transfers or controlled access. “Survival isn’t distribution” is the article’s decisive insight—and the fact its optimism cannot survive.