The cycle that began with [YOU] on AI asks who captures the gains of the AI transition and who bears its costs. Chang is the thinker who answers that question with the most empirical muscle. His framework insists that the question is not primarily about technology or markets—it is about who writes the rules and who has already climbed the ladder that the rules then prohibit others from using. The developer in Lagos that [YOU] on AI celebrates—the one now wielding tools once available only to well-funded teams—operates inside an institutional environment that Chang can map with historical precision: reliable electricity, payment infrastructure, intellectual property protections, domestic market access, and trade agreements in which Lagos had no voice. These are not natural features of the landscape. They are built. And the nations that built them built them the same way.
Chang sharpens the observation that [YOU] on AI makes about the orange pill requiring institutional dams to work. A dam in Chang's vocabulary has engineering specifications: tariff schedules, public education budgets, directed credit through state banks, technology transfer requirements, and the political will to resist pressure from the nations already at the top of the ladder. The Washington Consensus—the package of market liberalization that the IMF and World Bank pressed on developing nations through the 1980s and 1990s—was, in his reading, the ladder kicked away in policy form. Countries that followed it lost decades. Countries that defied it—China, Vietnam, South Korea earlier—are the growth stories of the era.
The AI transition reproduces this structural problem in a new domain. A handful of private companies, headquartered in one country, are making decisions—about training data, language priorities, pricing, and optimization targets—that constitute a global corporate industrial policy. No government voted for it. No developing nation negotiated its terms. The decisions are embedded in the technology itself and presented as engineering choices. Chang's framework names them as what they are: distributions of advantage, dressed as neutral outcomes.
His 2025 warning that India would be “one of the biggest casualties of AI”—because its growth model rested on exactly the service-sector activities that large language models automate most effectively—is the kind of historically grounded, uncomfortable prediction that the discourse tends to avoid. It is also the kind of prediction that history, given time, tends to confirm.
Born in Seoul in 1963, Chang grew up inside one of the most dramatic examples of his own thesis: the South Korean industrial miracle that Korean industrial policy built in a single generation. He studied economics at Seoul National University before earning his doctorate at Cambridge, where he came under the influence of heterodox economics—the tradition that takes seriously what mainstream theory excludes: history, institutions, power, and the gap between formal models and actual outcomes. His 2002 book Kicking Away the Ladder established his reputation and its central claim in one title: wealthy nations use the multilateral system to prohibit for others the policies they used themselves. 23 Things They Don't Tell You About Capitalism (2010) extended the argument into a systematic demolition of mainstream economic pieties, and Economics: The User's Guide (2014) offered the diagnostic toolkit to a general audience.
The cross-disciplinary method matters. Chang is an economist who reads history the way a prosecutor reads evidence—looking for the pattern that the official story conceals. When he observes that the United States maintained average tariffs above forty percent on manufactured imports during its entire century of industrialization, he is not making an ideological argument. He is presenting a fact that the standard account of American economic history suppresses, and drawing the obvious inference: that the free-trade advice now dispensed to developing nations is precisely the policy the United States could not afford to follow when it was in their position.
This method, applied to artificial intelligence, produces observations that the technology industry finds uncomfortable and that the historical record makes difficult to dispute. The internet was developed with public money, through DARPA. The algorithms underlying modern deep learning were developed at publicly funded universities. The semiconductor supply chains that manufacture AI chips were built through decades of industrial policy in Taiwan and South Korea. Chang is not arguing that the entrepreneurs who built Anthropic or OpenAI are not brilliant. He is arguing that their brilliance, exercised in a publicly funded ecosystem, does not justify the claim that the ecosystem is a product of private initiative—or that the public has no legitimate claim on how it develops.
Kicking Away the Ladder. The signature metaphor names a structural pattern that repeats across three centuries of economic history: wealthy nations build their prosperity through infant industry protection, tariffs, subsidies, and strategic state investment—then, once at the top, advocate the opposite for everyone still climbing. The ladder is not merely removed; the nations at the top give lectures about the moral superiority of not using ladders. The AI version of this pattern is the advocacy of open innovation and competitive markets by the nations and companies that have already accumulated the data, talent, computational infrastructure, and brand recognition that make those markets anything but equal.
The Amnesia of the Advantaged. The amnesia of the advantaged is not ordinary forgetting. It is a structural, ideologically functional erasure of the means by which wealth was produced, replaced by a story in which success was always already there—a natural consequence of superior institutions, superior values, superior people. The American AI industry's self-presentation as a triumph of private enterprise, operating in free markets, is the most recent and most brazen iteration of this amnesia. The publicly funded origins of the internet, the algorithms, and the semiconductor supply chain are a matter of living memory. Their erasure from the industry's origin myth is not incidental. It is essential to the claim that the industry deserves to capture all the returns.
Corporate Industrial Policy. The leading AI companies are making industrial policy decisions on a global scale—choices about which languages to optimize, which problems to prioritize, how to price access, which capabilities to develop and which to defer. These decisions have consequences as significant as any government tariff schedule, but they are made by private actors accountable to shareholders rather than democratic publics. The developing world needs development strategies, not compliance frameworks. A compliance framework tells Nigeria how to regulate AI. A development strategy tells Nigeria how to build an AI ecosystem. The difference between the two is the difference between consuming a technology and producing one.
Training Data as Enclosure. Chang's analysis of enclosure—the appropriation of the commons that preceded the English industrial revolution—maps directly onto the appropriation of collectively produced intellectual resources as AI training data. The programmer's open-source code, the writer's published text, the photographer's Creative Commons images: each was produced for a commons and scraped, without payment or consent, to train commercial systems worth hundreds of billions of dollars. The enclosure of the training commons is the largest involuntary transfer of intellectual value in economic history—and it is presented not as extraction but as progress.
Rules Matter More Than Technology. The technology is a given; it will advance regardless of what any government does. The rules determine who benefits from the advance. The rules of the AI ecosystem—intellectual property protections, safety standards, evaluation benchmarks, pricing structures—are being written now, by the actors with the most resources and the greatest stake in the current distribution of advantage. Chang's entire body of work is an argument that the seat at the rule-writing table must be claimed, not requested, and that claiming it requires understanding the rules well enough to know which ones to fight.