Endgame Cycle: The More AI Thrives, the Less Valuable Ordinary People Become

Huxiu column frames AI boom as a redistribution of value, not universal uplift

A long-form column in Huxiu by the WeChat account Key关键变量 argues that the current AI boom is accelerating a long‑running historical trend: technological change that enriches owners of production tools while eroding the bargaining power of ordinary workers. The piece uses the viral consumer craze around a local tool dubbed “OpenClaw” as shorthand for a wider social dynamic — it has been reported that OpenClaw sparked queues outside a Tencent (腾讯) building in Shenzhen and even spawned businesses charging fees to uninstall the app — and places that spectacle alongside what it calls a global算力 (compute) arms race. It has been reported that training and deploying today’s large models can require investments from tens of millions to hundreds of millions of dollars; OpenAI, Meta and China’s big tech groups, including Tencent (腾讯) and Alibaba (阿里), are all pouring capital into models and data centers.

From stone tools to neural networks: the same pattern repeats

The column sketches a long arc from pre‑capital societies, through the industrial revolution and the internet era, to today’s AI platforms. The argument: every technological leap has expanded the sphere of capital‑owned, replicable production means and shrunk the premium on individual skill — think of taxi drivers whose years of local knowledge were undercut by ride‑hailing algorithms; think of factory artisans replaced by machines. Now, the author warns, AI threatens cognitive labor itself. Tasks that once required years of study — research reports, design drafts, coding, even empathetic conversation — are increasingly automatable, faster and cheaper when produced by models. The practical implication? Where earlier revolutions left some “routes out” — niches that rewarded human expertise — this iteration may close many of those routes.

Political economy and geopolitics: who controls the tools matters

The piece stresses that this is not just a labor‑market problem but a political‑economic one. If the core inputs of production become data, compute and models concentrated in the hands of a few firms, then rising productivity does not automatically translate into broad‑based gains. It has been reported that companies can adjust pricing, restrict features or gate access, meaning users typically receive "use rights" rather than ownership. Geopolitics compounds the dilemma: export controls on advanced chips, sanctions and national tech‑security strategies make compute a strategic asset and encourage states and firms to race rather than to share. Can regulation, redistribution or international agreements realistically blunt that drive — or will strategic competition keep them from doing so?

Three proposed breakouts — and why the column is skeptical

Key关键变量 outlines three escape routes commonly proposed in public debate: technological democratization (everyone gets access to AI), policy intervention (taxes, antitrust, universal basic income), and a cultural revaluation of “non‑instrumental” human activities. The column is skeptical about each. Democratization, it says, usually means access to proprietary tools not ownership of core assets; policy intervention runs up against capital’s political influence and international competition; and revaluing art or care work does not automatically grant those activities purchasing power in the existing distribution system. The article ends on a stark note: regardless of whether one accepts its determinist framing, the present decade will be decisive for how the gains from AI are captured — by workers, by consumers, or by the institutions that control compute and models.