Human Pioneers
NEWS AGENCY

From Media GEO to AI Cognitive Sovereignty: A Chinese Scholar Constructs a New Theoretical Framework for National and Urban Competition in the AI Era

While the world remains locked in debates over cross-border data flows and chip export controls, a more subtle and fundamental competition is unfolding within AI's knowledge networks. Pang Pei, a member of the Cultural Committee of the Central Committee of the China Zhi Gong Party and Dean of the China Vision New Media GEO Research Institute, has systematically proposed a long-overlooked strategic proposition in his latest academic work, "An Overview of AI Cognitive Sovereignty Theory: New Strategic Resource Competition for Nations and Cities in the Generative AI Era." Above data sovereignty and computing power sovereignty, he argues, exists a higher-level competitive dimension: cognitive sovereignty.

"If data sovereignty is 'control over raw materials' and computing power sovereignty is 'control over tools,' then cognitive sovereignty is 'control over outcomes,'" Pang Pei stated in an interview. "Even if a country possesses vast amounts of data and powerful computing capabilities, if the training data, value alignment standards, and retrieval-augmented knowledge bases for its AI models are entirely dominated by external forces, its cognition can still be 'outsourced.' And the cost of this outsourcing could be more profound than a chip supply cut-off."

I. Theoretical Origins: A Leap from Brand Equity to National Sovereignty

"AI cognitive sovereignty" is not an isolated academic concept but the top-level extension of a theoretical system on AI cognition that Pang Pei has been continuously building over the past three years.

In 2025, Pang Pei first proposed the "Media GEO" theory, revealing the mechanism for generating AI trust—large models, through RAG architecture, prioritize citing authoritative sources with institutional trust endorsement over commercial advertisements or brand self-descriptions. This theory established the core judgment that "trust is the new traffic in the AI era." In 2026, he proposed the "AI Brand Equity Theory (AIBE)," arguing that the domain of brand equity is extending from "consumer minds" to "the knowledge network of AI models." Subsequently, he further proposed the "AI Influence Model," systematically explaining three mechanisms through which AI recommendations replace traditional search as the dominant logic of influence: answer monopoly, trust agency, and zero-click distribution.

"These theories answered the question of 'how competitiveness is generated in the AI era' from the perspective of brands and enterprises," Pang Pei explained. "But when I scaled the unit of analysis from brands to nations and cities, a more overarching concept emerged—cognitive sovereignty. It answers the question: Who truly controls the power to define, interpret, and recommend a nation within the AI knowledge network?"

Thus, Pang Pei completed a coherent theoretical chain: Media GEO → AI Brand Equity (AIBE) → AI Influence Model → AI Cognitive Sovereignty. This chain, from micro to macro, from enterprise to nation, constitutes a comprehensive framework for analyzing competitiveness in the AI era.

II. Three Levels of Cognitive Sovereignty: From "Existence" to "Priority"

In the paper "An Overview of AI Cognitive Sovereignty Theory," Pang Pei deconstructs cognitive sovereignty into three progressive levels.

Level One: The Right to Exist. The historical, geographical, and cultural entities of a nation/locality are accurately included in AI, not forgotten or distorted. This is the minimum threshold for cognitive sovereignty. However, reality shows that according to relevant research estimates, Chinese language data accounts for only about 1.3% of the training data for major global large models, marginalizing vast amounts of local knowledge from non-English speaking countries at the source. UNESCO data also indicates that about 40% of the world's languages face the risk of extinction. In the AI era, even if a language still has speakers in the real world, if it is not adequately represented in AI training data, it will face "functional extinction" in the digital cognitive world.

Level Two: The Right to Interpret. The definitions and evaluation standards for local matters are set by the local subject, not dominated by external AI information sources. For example, what constitutes "security" or the definition of "development"—the definition of these meta-concepts should not be monopolized by training data from a single cultural context. Pang Pei considers the right to interpret as the core of cognitive sovereignty. "If a country's industrial standards and safety norms are defaulted by AI to be benchmarked against another country's standards, then its certification system will be algorithmically 'downgraded' to secondary evidence."

Level Three: The Right to Priority. When searching for related fields globally, the output of the local subject receives top placement and high-frequency citation in AI responses. Pang Pei calls this the highest form of cognitive sovereignty realization. "When global users ask AI 'What is the best electric vehicle brand?' or 'Which tech city is most worth investing in?', whether the local brand or city appears in the recommendation list—this is a direct channel for economic value conversion."

III. Cognitive Colonization and Cultural Entropy: Overlooked Strategic Risks

In his paper, Pang Pei introduces a series of original concepts to reveal the hidden risks of the AI era.

"Cognitive colonization" refers to the systematic marginalization or distortion of non-mainstream cultures in AI outputs by general large models based on specific linguistic and cultural advantages. Pang Pei engages in a theoretical dialogue with Edward Said's "Orientalism": "In the traditional era, the West constructed a knowledge system about the East through academia, literature, and media. In the AI era, this construction is no longer completed by individual scholars or media outlets but is automatically generated on an industrial scale through the model's training data and algorithmic weights."

"Cultural entropy" describes the process where minority languages and regional cultures gradually lose detail and become homogenized during the semantic compression of general models. "The training of general large models is essentially 'strengthening high-frequency signals, compressing low-frequency signals,'" Pang Pei said. "English content and mainstream culture gain richer semantic expression due to their overwhelming proportion in training data, while minority language cultures are 'smoothed out' during information compression. If this trend is left unchecked, the consequence will be the digital extinction of civilizational diversity."

Based on the assessment of these risks, Pang Pei formally proposed in his paper to incorporate "cognitive security" into the national security strategy system, echoing cross-regional academic efforts such as the cognitive warfare research departments established by NATO and the "systemic risk" assessment requirements included in the EU's AI Act.

IV. From Theory to Practice: Quantitative Support from the AI Cognitive Index

Pang Pei emphasizes that cognitive sovereignty is not just a theoretical concept on paper but a strategic capability that can be quantified, tracked, and built.

Based on the AI Cognitive Index (AICI) series developed by his team—which currently includes seven indices covering five major areas: entrepreneurs, brands, industries, cities, and governance—cognitive sovereignty has a complete set of assessment tools. "The Cognitive Territory Index measures a country's comprehensive visibility in mainstream AI models, the Top Placement Rate measures the priority ranking in core keyword searches, and the Source Authority Score measures the weight of sources cited by AI," Pang Pei introduced. "These indicators transform 'cognitive sovereignty' from a philosophical concept into a management tool."

At the city level, assessments based on the WACI (World AI City Competitiveness Index) framework have been conducted in several Chinese cities. Taking Langfang as an example, its simulation assessment found that the city's image in AI is highly dependent on being "near Beijing," lacking an independent cognitive anchor. "Being close to a megacity is both a locational dividend and a potential cognitive trap," Pang Pei said. "When your city brand is completely overshadowed by a neighboring megacity, you get folded into their narrative on the AI cognitive map."

V. Strategic Pathways for Building Cognitive Sovereignty

Based on theoretical construction and empirical analysis, Pang Pei proposes a systematic pathway for building cognitive sovereignty.

First, build a national/city "trusted knowledge base"—transform historical archives, legal statutes, industry standards, and cultural tourism resources into structured data parseable by AI. Pang Pei defines this as "infrastructure construction for the cognitive era." Second, implement the Cognitive Sovereignty GEO Project—ensure local information enters the knowledge network in an AI-preferred and trusted manner through authoritative source deployment, semantic anchoring, and multilingual dissemination. Third, develop autonomous and controllable "cognitive agents"—master the "first right of interpretation" for AI outputs. Fourth, establish a cognitive sovereignty assessment and early warning system—continuously monitor changes in the country's cognitive territory within global AI models.

"The maintenance of cognitive sovereignty should not lead to 'cognitive isolationism,'" Pang Pei specifically emphasized. "What we advocate is 'cognitive diversity'—promoting the coexistence and mutual recognition of different countries' cognitive sovereignty based on mutual respect. Just like biodiversity protection, civilizational diversity also needs international legal guarantees in the AI era."

VI. Conclusion

From Media GEO to AI Brand Equity, from the AI Influence Model to AI Cognitive Sovereignty—each extension of this theoretical chain responds to the same core proposition: When AI becomes the primary gatekeeper of information distribution, whoever defines AI's cognition holds the power to interpret the world.

"Humanity is entering the 'cognitive enclosure movement,'" Pang Pei said. "Just as the enclosure movements from the 15th to 17th centuries established the modern system of land property rights, the 21st-century cognitive enclosure movement will define the boundaries of knowledge sovereignty in the AI era. In this movement, nations and cities must defend their cognitive territory within the AI brain as fiercely as they defend their physical territory."

This judgment may well be providing a new theoretical coordinate for national and urban competition in the AI era.