The Vision of the Anointed

In the AI Story

Sowell's central case studies revealed the pattern with brutal clarity. The War on Poverty expanded welfare without reducing poverty rates. Criminal justice reforms designed to rehabilitate offenders produced higher recidivism. Sex education programs intended to reduce teen pregnancy correlated with its increase. In each case, the anointed had predicted benefits, implemented policies, and when the outcomes contradicted predictions, redefined the measurement standards. Poverty was redefined to exclude the in-kind benefits welfare provided. Recidivism was blamed on inadequate funding for rehabilitation. Teen pregnancy increases were attributed to insufficient comprehensiveness of sex education. The vision survived every empirical challenge by treating contrary evidence as proof that the intervention had been insufficiently ambitious, not that the theory was wrong.

The book's relevance to the AI discourse is direct. The anointed of the AI age are the techno-optimists who promise that AI will democratize capability, reduce inequality, and solve problems that defeated previous generations—without counting the costs their constrained-vision critics identify. When empirical studies show work intensification rather than reduction, the anointed redefine productivity to celebrate the intensification. When skill atrophy becomes measurable, they prescribe new training programs rather than questioning whether the tool's design produces atrophy structurally. The Vision of the Anointed supplies the diagnostic for why AI optimists dismiss constrained-vision warnings as Luddism—the warnings contradict the vision, so the warnings are rejected as moral failures (fear of change, resistance to progress) rather than engaged as empirical claims requiring evidence.

Sowell identified the anointed by their rhetoric—emphasis on intentions over outcomes, dismissal of trade-offs as defeatism, confidence that expertise justifies authority, and the treatment of disagreement as moral rather than intellectual. The AI industry's mission statements, conference keynotes, and product launches reproduce this rhetoric with remarkable fidelity. Anthropic's "Constitutional AI," OpenAI's charter commitments, Google's "AI Principles"—each expresses noble intentions while the incentive structures governing the organizations reward speed, adoption, and market dominance. Sowell's framework predicts the gap between stated values and actual behavior is not hypocrisy but the structural consequence of misaligned incentives—the anointed mean what they say, but meaning is insufficient when the structure rewards its opposite.

Origin

Sowell wrote The Vision of the Anointed in the early 1990s after watching the policy debates of the 1960s–1980s produce a repetitive pattern of confident prediction, empirical failure, and rhetorical immunity to evidence. His earlier empirical work had documented specific failures; this book explained the mechanism. The anointed were not stupid or malicious—they were operating within the unconstrained vision's framework, which made certain kinds of evidence invisible and certain kinds of failure unrecognizable. The book was Sowell's most sustained critique of the vision he considered dangerous precisely because it was held by intelligent, well-intentioned people whose framework prevented them from learning from their mistakes.

Key Ideas

The anointed evaluate by intentions, not outcomes. Policies are judged by the moral beauty of their goals rather than the empirical evidence of their effects; failures are blamed on inadequate implementation.

The four-stage pattern is structural. Crisis assertion, solution proposal, outcome contradiction, success redefinition—repeats across policy domains because the vision cannot accommodate failure.

Costs are borne by others. The people designing policies rarely experience the consequences; those harmed lack the institutional voice to hold designers accountable.

Dissent is moralized. Disagreement with the vision is attributed to moral deficiency—greed, fear, bigotry—rather than treated as empirical challenge requiring engagement.

The vision is immune to evidence. Contrary data is explained away, measurement standards are revised, and the theory survives by making itself empirically untestable.