Zheng Yongnian

In the [YOU] on AI Field Guide

[YOU] on AI describes the orange pill as the moment when a person genuinely sees what AI is doing to the structure of cognition, expertise, and civilization’s self-reproduction. Zheng’s contribution to the cycle is to take that pill at the civilizational level and report what is visible from the other side. What he sees is a technology that does not merely disrupt existing institutions but actively reshapes the civilizational conditions under which those institutions operate—and that, because AI systems are trained on and optimized for particular cultural assumptions, the reshaping is not neutral. A civilization that governs AI with borrowed frameworks is not protecting its own values; it is allowing a technology designed elsewhere to overwrite them.

His argument that American AI development resembles a powerful engine without adequate brakes—prodigious frontier capabilities, weak governance—while Chinese AI development resembles an engine with brakes but insufficient horsepower, situates the Sino-American AI competition not as a race to the same finish line but as two asymmetric civilizational experiments that the world will learn from in different ways. The cycle’s insistence that AI is an amplifier whose quality depends on the quality of what is amplified finds its civilizational extension in Zheng: the amplifier will amplify the values embedded in its design, and those values belong to the civilization that built it.

He is the cycle’s most important interlocutor on the non-Western stakes of AI governance, and his framework complements the individual-level analyses of Yuval Noah Harari and the attentional-ecology work of Yves Citton by operating at the level where those analyses converge: the question of what kind of civilization—not just what kind of individual—can survive the AI transition with its self-understanding intact.

Origin

Zheng Yongnian was born in Zhejiang province and trained in political science in China before receiving his doctorate from Princeton. He has been a professor at the National University of Singapore and is currently founding director of the Advanced Institute for Global and Contemporary China Studies at the Chinese University of Hong Kong, Shenzhen. His scholarly reputation rests on his analysis of the Chinese Communist Party as an organizational institution with deep continuity with the imperial tradition—a structural observation that resisted the dominant Western frame of China as a failed liberal democracy in progress.

His 2010 analysis of the Chinese state, The Chinese Communist Party as Organizational Emperor, established his signature analytical move: understanding Chinese institutions on their own terms rather than as deviations from Western models. Technological Empowerment (2008) applied this lens to the internet in China, arguing that the state was not merely censoring the internet but co-evolving with it—shaping what the internet became within the Chinese civilizational context in ways that Western platform analysis systematically missed.

His engagement with AI governance began in earnest around 2024–2025, culminating in a paper in the Bulletin of the Chinese Academy of Sciences that identified the structural characteristics of AI systems—highly concentrated capabilities, centralized control, monopolized deployment, diffused users—as inherently dangerous governance problems requiring a civilizational-level response. His concept of artificial ignorance as the AI-era counterpart to previous industrial revolutions’ liberation of physical labor introduced a distinctive contribution: not merely that AI displaces workers, but that it displaces the cognitive activity on which social order depends.

Key Ideas

The civilizational state. China is not a strong version of a familiar state type but a different type entirely: a civilizational state whose claim to legitimacy rests on performance—the capacity to deliver stability, prosperity, and good governance in the Confucian sense—rather than on procedural mandate. This framing transforms the analysis of Chinese AI governance: the state intervenes not because it is uniquely controlling but because its legitimation logic requires intervention when a technology produces outcomes that damage social order. The performance-based state must govern for outcomes; it cannot satisfy its mandate by following the correct procedure while society deteriorates.

Against intellectual colonialism in AI ethics. The dominant AI ethics frameworks emerged from particular cultural contexts and are not portable without distortion. An AI ethics built on individual rights, transparency to individuals, and markets as the primary allocation mechanism reflects the liberal-democratic tradition and is not adequate to governance in a Confucian civilizational context that begins from the relational self, social harmony, and state capacity as the condition of civilization rather than its threat. Applying Western frameworks to non-Western governance is not universal ethics; it is the imposition of one civilization’s self-understanding on another’s self-organization.

Artificial ignorance and de-intellectualization. The three previous industrial revolutions liberated humans from physical and routine cognitive labor, expanding the scope for thinking. AI may for the first time liberate humans from intellectual labor itself, producing what Zheng calls artificial ignorance: not individual stupidity but collective de-intellectualization, the progressive atrophy of the cognitive capacities that a social order requires for its own reproduction. A society that outsources its thinking cannot govern the technology it has outsourced to—and a social order that cannot think cannot sustain itself.

The pastoral society model. Zheng’s evocative term for the AI-enabled governance dynamic in which a small number of entities—platform companies, state agencies—play the role of the shepherd, organizing the movement of a passive human population that has surrendered its navigational capacity to the herder. The metaphor names a structural dependency: when cognitive navigation is concentrated in one entity and a large population becomes habituated to relying on it, the distributed navigational capacity atrophies, and the population loses not just the practice but the faculty of finding its own way.

The multipolar AI world. Sino-American AI competition is not a race to the same finish line but a contest between two civilizational traditions with different strengths: American AI as an engine without adequate brakes (frontier capability, weak governance), Chinese AI as an engine with brakes but insufficient horsepower (governance capacity, constrained frontier). The outcome that serves the world is not the dominance of either model but a multipolar AI governance architecture adequate to the diversity of civilizational traditions—one that requires, as its precondition, each tradition doing the difficult work of civilizational self-examination rather than borrowing frameworks developed elsewhere.