Two visions of society: Reading Thomas Sowell (托马斯·索维尔)

Sowell’s divide: the “anointed” vs the “tragic”

A short essay on Huxiu (虎嗅) revisits Thomas Sowell’s The Intellectuals and Society (知识分子与社会) and foregrounds a question that matters for today’s tech debate: are we living under the logic of the “vision of the anointed” or the “tragic vision”? Thomas Sowell (托马斯·索维尔) casts intellectuals as producers of ideas who lack the same real‑world feedback as practitioners. That distance, he argues, can make some thinkers overconfident about sweeping, plan‑based remedies for social ills.

Sowell’s “anointed” see social problems as correctable by grand designs — new theories, revolutionary institutions, or central plans that will finally right injustices. The “tragic” vision is humbler: human nature and institutions are imperfect; progress depends on accumulated experience, cautious reforms and protections like property rights. The Huxiu writer uses Sowell to diagnose contemporary debates about planning, expertise, and moral certainty.

Why it matters for AI and tech optimism

Why does a 20th‑century social critique matter to China’s tech scene and to global AI debates? Because it maps neatly onto two competing attitudes toward emergent technologies. On one side are the techno‑utopians who portray general artificial intelligence as a near‑miraculous solver of systemic problems — governance, health, inequality. It has been reported that some founders and commentators in both China and the West frame AI as transformative enough to substitute for slow, contested political compromises. On the other side are those who warn that powerful tech must be constrained by institutions and distributed knowledge — the tragic skeptics who stress unintended consequences, governance complexity and incremental learning.

This is also a geopolitical issue. Debates about AI governance, export controls and tech decoupling between the U.S. and China turn on similar assumptions: can a bold set of policies or platforms reliably engineer better social outcomes, or do such efforts risk magnifying error at scale? Reportedly, regulators on both sides are grappling with these questions even as companies race to commercialize advanced models.

Which side are today’s AI evangelists on? If you believe in a single revolutionary architecture that will cure human problems, you resemble the anointed. If you believe civilization is a fragile achievement that requires institutions, modesty and iterative learning, you inhabit the tragic vision. The Huxiu piece — and Sowell’s argument — is a timely reminder: technological ambition needs to be paired with epistemic humility.