Language：
```html
World AI City Competitiveness Index (WACI) White Paper
```World AI City Competitiveness Index (WACI) White Paper
(2026 Edition)
The Urban Cognitive Competitiveness Revolution in the AI Era
— From City Strength Competition to AI Cognitive Rights Competition
Compiled by:
- World Intelligence Organization (WIO) / AI Index Research Institute (A global non-profit organization registered with the UN Economic and Social Council, holding U.S. federal 501(c)(3) tax-exempt status)
- GEO Research Institute of CCTV New Screen Internet TV Co., Ltd. (A company invested by Central Newsreel and Documentary Film Studio Group under China Media Group)
- WICOWEB Global AI Cognitive Research Center
- Beijing North Future Vocational Skills Appraisal Center (Holding a nationwide foreign-related survey license issued by the National Bureau of Statistics)
Theoretical Support: PAI Framework (Pangpei AI Index Framework)
Lead Proponent: Pei Pang
August 2026
The Urban Cognitive Competitiveness Revolution in the AI Era — World AI City Competitiveness Index (WACI) White Paper
About This White Paper
This white paper is compiled based on the Pangpei Index · World AI City Competitiveness Index (WACI 2.0) evaluation framework. The data referenced herein derives from a comprehensive analysis combining simulated assessments and preliminary real-world data collection. Some cities have undergone multiple rounds of data validation, while data for other cities is still being continuously refined. This white paper aims to elaborate on the theoretical framework and evaluation methodology of WACI, present trend-based findings based on available data, and provide city administrators and researchers with conceptual tools and strategic references for building urban competitiveness in the AI era.
WACI's global city assessment efforts will continue to advance. As the data collection system is continuously improved and the city sample gradually expands, a more comprehensive and systematic official version of the white paper will be released in the future.
Abstract
As global capital and talent increasingly rely on AI large models to make decisions regarding investment location selection, career migration, and business activities, a new dimension of urban competitiveness is emerging — AI cognitive competitiveness. The World AI City Competitiveness Index (WACI) is the world's first urban AI cognitive competitiveness assessment system that uses real output data from AI large models as its evaluation basis. It systematically measures a city's "presence," "distinctiveness," and "recommendation priority" in the AI knowledge network from four dimensions: cognitive visibility, industry labeling power, recommendation advantage, and cognitive resilience. Based on the WACI evaluation framework, this white paper presents for the first time a systematic overview of the competitiveness landscape of approximately 60 Chinese cities in the AI cognitive world, reveals three core laws of urban competition in the AI era, and proposes a systematic pathway for building urban AI cognitive competitiveness.
Table of Contents
Preface: When AI Begins Choosing Cities for Humans
Chapter 1: Urban Competition Enters the "Cognitive Era"
1.1 Three Transitions in the Competitive Paradigm
1.2 Cognitive Competitiveness: A Neglected Strategic Dimension
1.3 From GDP Rankings to AI Rankings
Chapter 2: WACI — The Urban Coordinate System in the AI Era
2.1 Evaluation Philosophy and Core Principles
2.2 The Four Core Dimensions of WACI
2.3 Data Collection and Processing Methods
Chapter 3: Preliminary Findings — Chinese Cities Through the Lens of AI
3.1 The Landscape of AI Cognitive Competitiveness in Chinese Cities
3.2 Key Finding 1: Label Clarity Matters More Than Economic Scale
3.3 Key Finding 2: The "Cognitive Parasitism Effect" — A New Moat in the AI Era
3.4 Key Finding 3: The Risk of "Cognitive Collapse" Deserves Attention
3.5 Key Finding 4: Distinctive Cities Build Unique Cognitive Advantages on Differentiated Tracks
3.6 The "Language Gap" Phenomenon in Multilingual Coverage
Chapter 4: Three Core Laws of Urban Competition in the AI Era
4.1 Law 1: From "Scale Competition" to "Label Competition"
4.2 Law 2: From "Event-Driven" to "Asset Building"
4.3 Rule Three: From "Passive Waiting" to "Active Construction"
Chapter 5: City Brand Building Enters the Era of AI Cognitive Operations
5.1 Limitations of Traditional City Brand Building
5.2 Four Transformations in City Brand Building
5.3 AI Cognitive Assets: The Most Important New Asset for Future Cities
Chapter 6: Action Guide — The Construction Path for Urban AI Cognitive Competitiveness
6.1 Step One: Cognitive Diagnosis
6.3 Step Three: Authoritative Source Layout
6.4 Step 4: Cognitive Resilience Building
6.5 Step 5: Multilingual Coverage
Chapter 7: From China to the World — WACI's Global Vision
Conclusion: Whoever Masters AI Cognitive Rights Defines the Future Competitiveness of Cities
Appendix A: Complete Description of the WACI 2.0 Indicator System
A.2 Assessment Dimensions and Weights
A.3 Sub-Dimensions and Basic Indicators
A.4 Data Collection and Processing
Appendix: Original Text of the WACI 2.0 Index Model
III. Assessment Dimensions and Indicator System
IV. Data Collection and Calculation
VI. Publication and Application
Appendix B: List of Assessed Cities
B.4 Southeast Asia and South Asia (25 cities)
B.5 Middle East and Africa (25 cities)
B.8 City Sample Management Standards
Appendix C: Standardized Question Set (Example)
C.2 Cognitive Visibility Issues
C.4 Recommendation Advantage Issues
C.5 Fixed Tracking Issues (Core Question Set, 30 Questions in Total)
C.6 Question Set Management Specifications
Appendix D: Authoritative Source Library (Excerpt)
D.1 Source Classification Standards
D.2 S-Level Sources: Global Institutional Sources
D.3 A-Level Sources: National-Level Authoritative Sources
D.4 Tier B Sources: Industry Authority Sources
D.5 Source Library Management Standards
D.6 Complete Source Library Description
Appendix E: Data Sources and Disclaimer
Foreword: When AI Begins Choosing Cities for Humans
In 2026, human society is undergoing a transformation more profound than the internet. Over the past two decades, people grew accustomed to finding answers through search engines; today, more and more people are turning directly to artificial intelligence with their questions.
"Which Chinese cities are the most worth investing in for the future?"
"Which city is best suited for the development of the AI industry?"
"Where is the most innovative city in Asia?"
"Which city should young entrepreneurs choose to start a business?"
The changes behind these questions go far beyond an upgrade in how information is accessed. Because AI is no longer just a search tool—it is becoming a decision advisor, a cognitive assistant, and a recommendation engine.
In the past, urban competition primarily took place in the physical world. Companies evaluating cities needed research reports, talent choosing cities needed firsthand comparisons, and investment institutions seeking opportunities required long-term study. Today, these processes are increasingly being participated in or even restructured by AI. AI is helping people complete information gathering, material screening, plan comparison, and even preliminary decision recommendations. Data shows that by mid-2026, the global monthly active users of generative AI have surpassed 2 billion, and nearly 70% of consumers make purchasing decisions based on recommendations output by AI platforms. Similar trends are accelerating in the fields of investment site selection, talent mobility, and industrial layout.
This means that whether a city can be accurately understood, consistently cited, and preferentially recommended by AI will directly affect future talent flows, capital flows, industrial flows, and attention flows. A new form of urban competition is taking shape: not competing for land, not competing for population, not competing for traffic, but competing for AI cognitive rights—the right of existence, the right of interpretation, and the right of recommendation for a city within the AI knowledge network.
It is against this backdrop that the World AI City Competitiveness Index (WACI) has emerged. It seeks to answer an unprecedented question: in the eyes of artificial intelligence, where exactly do the world's major cities stand?
Chapter One: Urban Competition Enters the "Cognitive Era"
1.1 Three Shifts in the Competitive Paradigm
Looking at the history of urban development, each era has its own core logic of competition.
In the agricultural era, cities competed for land and water resources;
In the industrial era, cities competed for industrial foundations and transportation hubs;
In the information era, cities competed for capital, talent, and internet traffic.
Pang Pei's AI Cognitive Competition Theory points out that the paradigm of human competition is undergoing a third shift:
The industrial era was one of "resource competition"—whoever possessed more natural resources and production capacity held the competitive advantage;
The internet era was one of "traffic competition"—whoever captured more user attention held the key to growth;
The AI era is one of "cognitive competition"—whoever holds greater cognitive weight, a more stable trust rating, and a higher recommendation priority within the AI knowledge network holds the "key" to the gateway of user cognition.
At the urban level, cognitive competition manifests as the degree to which a city is seen, trusted, and recommended within the AI knowledge network. The concept of "AI cognitive rights" proposed by Pang Pei—the right of existence (being accurately included by AI, not forgotten or distorted), the right of interpretation (defining one's own affairs and evaluation standards), and the right of priority (obtaining reasonable visibility and recommendation priority in global searches of relevant fields)—provides a systematic analytical framework for understanding a city's competitiveness in the AI era.
1.2 Cognitive Competitiveness: A Neglected Strategic Dimension
So-called cognitive competitiveness refers to an entity's ability to be understood, remembered, cited, and recommended. In the past, cognition primarily existed within the human brain; today, cognition is being massively deposited into AI models. More and more people are learning about the world through AI, more and more companies are finding partners through AI, and more and more investment institutions are conducting industry research through AI. AI is becoming an important gateway for humanity to understand the world.
Therefore, if a city cannot enter the AI cognitive system, it means it is losing an important gateway to the future. Competition among cities in the future will not only involve vying for the allocation of resources in the physical world, but also for discourse power and recommendation rights in the AI world.
1.3 From GDP Rankings to AI Rankings
For a long time, evaluating cities has mainly relied on indicators such as GDP, fiscal revenue, industrial scale, and population size. These indicators can reflect a city's level of development, but they cannot explain a growing number of new phenomena:
Why do some cities, whose GDP is not the highest, frequently appear on lists of recommended global innovation cities?
Why do some cities, whose population size is not dominant, become AI's top recommended cities for entrepreneurship?
Why do some cities with similar industrial scales show enormous gaps in global recognition?
The reason lies in the fact that real-world strength and cognitive influence are not entirely equivalent. A city may possess a strong industrial base yet lack global recognition; a city may be limited in size yet wield an extremely powerful label effect. In the AI era, cognition is becoming a new form of productivity. A city's influence depends not only on what it "is," but also on what AI perceives it to "be." Therefore, the future urban evaluation system needs to add a new dimension: AI cognitive competitiveness. WACI is a new evaluation system established precisely to address this need.
Chapter 2: WACI — The Urban Coordinate System of the AI Era
2.1 Evaluation Philosophy and Core Principles
WACI does not evaluate a city's GDP, fiscal revenue, housing prices, or population size. What WACI focuses on is a city's position within the global AI knowledge system. Its core questions are:
Does AI recognize this city?
How does AI describe this city?
Is AI willing to recommend this city?
Does AI trust this city?
Can AI accurately understand this city's industrial advantages and development potential?
The core philosophy of WACI is: it is not people evaluating cities, but how AI "perceives" and "recommends" cities. This philosophy is built upon four major principles:
Quantifiable data — all evaluation indicators are based on quantitative analysis of AI's actual output data;
Reproducible methodology — standardized collection processes ensure third-party independent verification;
Cross-model comparability — multi-model, multi-language cross-validation eliminates single-model bias;
Dynamic traceability — annual collection combined with quarterly tracking captures temporal changes.
The theoretical foundation of WACI originates from the PAI Framework proposed by Pang Pei, as well as media-based GEO theory, AI Brand Equity (AIBE) theory, and the AI Influence Model:
Media-based GEO theory: reveals why AI trusts authoritative sources over advertising — this explains why the density of a city's presence in authoritative sources directly determines its cognitive weight in AI.
AI Brand Equity theory: constructs a dimensional framework for building a city brand's cognitive assets in the AI era.
AI Influence Model: demonstrates the deep mechanisms by which AI recommendations become the dominant logic of influence — answer monopoly, trust agency, and zero-click distribution.
Together, these theories point to one core judgment: the future of competition is no longer just about being seen by people, but being seen by AI; no longer just about being remembered by people, but being remembered by AI.
2.2 The Four Core Dimensions of WACI
Dimension 1: AI Cognitive Visibility (Weight 30%)
This is the first threshold for a city to enter the AI world. If AI cannot recognize a city, all subsequent competition is moot. Cognitive visibility primarily measures: the frequency with which a city is mentioned in mainstream AI models, the breadth of scenarios covered by information, multilingual communication capabilities, and presence in the global knowledge system. In the future, city official websites, industry reports, academic research, media coverage, and international cooperation outcomes will all become important sources of AI cognition. The higher the visibility, the easier it is for a city to enter the global AI knowledge network.
Dimension 2: AI Trustworthiness and Industrial Label Strength (Weight 30%)
AI must not only recognize a city, but also trust it. This dimension measures: the clarity and binding strength of AI's labels for a city's industries, the consistency of AI's descriptions of the city, information accuracy, the proportion of authoritative source citations, and the stability of positive perceptions. In the AI era, trust has become a new competitive asset. A city supported by authoritative sources with clear industrial labels is more likely to receive sustained recommendations from AI. Industrial label strength is the most differentiating indicator of this dimension — it answers the core question of "what AI believes this city is known for."
Dimension 3: AI Recommendation Rate (Weight 25%)
This is one of the most important indicators of WACI. Because the core force influencing the flow of resources in the future is not simple exposure, but recommendation. When AI answers questions like "which cities are most worth investing in," "which cities are most suitable for entrepreneurship," and "which cities have the greatest future growth potential," the recommendation ranking will directly affect real-world decision-making. Whoever is recommended first is more likely to gain attention; whoever gains more attention is more likely to gain more opportunities.
Dimension 4: AI Cognitive Resilience (Weight 15%)
A truly strong city brand does not exist in only one model, but forms consistent perceptions across multiple models. Whether it is GPT, Gemini, Claude, DeepSeek, or other large models, if they all arrive at similar conclusions, it indicates that the city has formed stable cognitive assets. Cognitive resilience also measures the stability and recovery speed of a city's AI image during model version updates and public opinion incidents. The stronger the cognitive resilience, the greater the long-term value of the city brand.
Table 1: WACI 2.0 Four-Dimensional Evaluation Framework
| Dimension | Weight | Core Proposition | Key Indicators |
| Cognitive Visibility | 30% | Does AI know this city? | Cross-model mention rate, cross-language mention rate, scenario coverage, language distribution balance |
| Industry Labeling Power | 30% | What does AI consider this city known for? | Dominant industry label binding strength, innovation label co-occurrence rate, industry description information density, benchmark enterprise/institution citation rate |
| Recommendation Advantage | 25% | Under what circumstances does AI recommend this city? | Top recommendation placement rate, recommendation scenario coverage, cross-model recommendation consistency |
| Cognitive Resilience | 15% | How stable is this city's AI cognitive asset? | Cross-model version stability, cross-model cognitive consistency, negative public opinion recovery cycle |
2.3 Data Collection and Processing Methods
WACI sends queries to mainstream global AI large models through standardized API interfaces, covering models such as the GPT series, Gemini, Claude, ERNIE Bot, DeepSeek, Tongyi Qianwen, Doubao, etc. The language coverage includes seven languages: Chinese, English, French, Spanish, Arabic, Japanese, and Korean, which together cover approximately 65% of the global population, over 80% of global GDP, and more than 90% of AI training corpora.
The city sample covers approximately 200 major cities worldwide, spanning more than 50 countries and regions across six continents. Each city is designed with approximately 50 standardized query questions, covering scenarios such as basic city awareness, economic status, technology positioning, industrial dominance, business site selection, talent migration, and investment recommendations. Each question is independently queried 3 times on each model (with intervals of no less than 24 hours), and the average is taken to reduce randomness in AI responses. Two rounds of data collection are conducted annually, supplemented by quarterly dynamic tracking.
Data processing employs multilingual named entity recognition (NER) to extract city names and associated entities, sentiment analysis to determine the emotional tendency of AI descriptions, industry label extraction and binding strength calculation to measure the cognitive anchoring of cities in specific tracks, source authority assessment to evaluate the quality level of AI-cited sources, and cross-language consistency calibration to eliminate language bias.
At the current stage, some cities have completed multiple rounds of systematic data collection, while data for some cities is based on comprehensive analysis of simulated assessments and initial collection. As the data collection system continues to advance, city coverage and evaluation accuracy will be continuously improved.
Chapter 3: Preliminary Findings — Chinese Cities in the Eyes of AI
3.1 Competitive Landscape of Chinese Cities in AI Cognition
Based on the WACI evaluation framework, approximately 60 Chinese cities present a clear hierarchical structure.
AI Tier-1 Cities (comprehensive score ≥90): Shenzhen, Beijing, Shanghai, Hangzhou. These four cities perform prominently in the AI cognitive world, each possessing distinct and stable cognitive labels — Shenzhen as "China's Silicon Valley," Beijing as "China's AI Brain," Shanghai as "Global Financial Center," and Hangzhou as "No.1 City in Digital Economy." These labels have been deeply embedded in the AI knowledge system. Shenzhen tops the list with the "China's Silicon Valley" label, and when AI answers "China's most innovative city," Shenzhen is almost inevitably the first choice. Beijing follows closely with the "China's AI Brain" label, and the clustering effect of AI enterprises such as DeepSeek, Baidu, and ByteDance positions Beijing as a core node in global AI industry cognition.
AI First-Tier Cities (comprehensive score 80-89): Hefei, Chengdu, Guangzhou, Wuhan, Nanjing, Suzhou, Xi'an, and Chongqing. These eight cities have formed strong AI recognition labels in their respective distinctive tracks. Hefei's "most impressive venture capital city" and quantum technology, Chengdu's "western technology hub," Wuhan's "Optics Valley," and Xi'an's "city of hard technology" — these cities are rapidly rising in the AI recognition network through their distinctive industrial advantages.
AI New First-Tier Cities (comprehensive score 70-79): Changsha, Tianjin, Wuxi, Zhengzhou, Dongguan, Qingdao, Ningbo, Xiamen, Jinan, Fuzhou, Zhuhai, and Foshan. These cities have competitive advantages in specific dimensions, but there is still room for improvement in their overall AI recognition weight.
AI Featured Cities (comprehensive score 60-69): 16 cities including Kunming, Guiyang, Shenyang, Dalian, and Harbin. Guiyang stands out in this tier with its "China's data valley" label, making it one of the cities with the strongest industrial label recognition among featured cities.
AI Potential Cities (comprehensive score 50-59): 12 cities including Huizhou, Jiaxing, Shaoxing, Wenzhou, and Quanzhou. These cities are in the early stages of AI recognition development but have the potential to break through in specific tracks.
AI Watch Cities (comprehensive score 40-49): 8 cities including Jinhua (Yiwu), Guilin, Lhasa, Xiong'an New Area, and Dunhuang. Although these cities have limited economic scale, they possess unique AI recognition presence in specific tracks — Yiwu's "world's capital of small commodities," Dunhuang's "world cultural heritage," and Xiong'an New Area's "city of the future."
Table 2: WACI tier distribution of Chinese cities
| Level | Score Range | Number of Cities | Representative Cities |
| AI Super First-Tier Cities | ≥90 points | 4 | Shenzhen, Beijing, Shanghai, Hangzhou |
| AI First-Tier Cities | 80-89 points | 8 | Hefei, Chengdu, Guangzhou, Wuhan, Nanjing, Suzhou, Xi'an, Chongqing |
| AI New First-Tier Cities | 70-79 points | 12 | Changsha, Tianjin, Wuxi, Zhengzhou, Dongguan, Qingdao, etc. |
| AI Featured Cities | 60-69 points | 16 | Guiyang, Kunming, Shenyang, Dalian, Harbin, etc. |
| AI Potential Cities | 50-59 points | 12 | Huizhou, Jiaxing, Shaoxing, Wenzhou, Quanzhou, etc. |
| AI Watch Cities | 40-49 points | 8 | Jinhua (Yiwu), Guilin, Lhasa, Xiong'an New Area, Dunhuang, etc. |
3.2 Key Finding 1: Label Clarity Matters More Than Economic Scale
The WACI study reveals a core pattern: when AI recommends cities, it prioritizes not economic scale but the clarity of industry labels.
Hefei and Guiyang are typical cases of this pattern:
Hefei does not rank at the top in traditional GDP rankings, yet it has secured a place among AI first-tier cities in the WACI. Its core driver is not GDP size but the deep integration of the highly communicable industry label "China's boldest venture capital city" within AI. The clustering effect of leading enterprises such as BYD, NIO, and CXMT, along with global recognition in quantum information research, makes Hefei far more likely to be recommended when AI answers questions about "China's technological innovation cities" compared to cities of similar GDP scale.
Guiyang is no exception. The "China Data Valley" label has given this western city an irreplaceable position in AI's industrial perception. When AI is asked about "China's big data industry center," Guiyang is almost always a mandatory answer.
This confirms a core judgment in Pang Pei's AI cognition theory: in the AI era, label clarity is becoming a more decisive variable for urban competitiveness than economic scale. A medium-sized city with an industry label that AI cannot bypass may achieve higher global recognition than a megacity with vague industry labels.
3.3 Key Finding 2: The "Cognitive Parasitism Effect" — A New Moat in the AI Era
During the WACI research, a highly valuable phenomenon was discovered — the cognitive parasitism effect. The so-called cognitive parasitism effect refers to a situation where, after a subject becomes highly bound to a certain field, AI can hardly avoid mentioning it when discussing that field. For example:
When mentioning new energy vehicles, AI will prioritize BYD and Shenzhen;
When mentioning drones, AI will prioritize DJI; when mentioning the digital economy, AI will prioritize Hangzhou;
When mentioning big data, AI will prioritize Guiyang.
At this point, these subjects have become part of the knowledge system in that field. For cities, once the cognitive parasitism effect is formed, it means an extremely high competitive barrier has been established. In Pang Pei's AI cognition theory, this effect stems from long-term, multi-source, and authoritative source accumulation — cities form "unavoidable" semantic anchors in the AI knowledge network through authoritative media coverage, industry white paper releases, academic paper citations, and participation in international standards. The key to future competition is not to become the best, but to become the answer that AI cannot bypass.
3.4 Key Finding 3: The Risk of "Cognitive Folding" Deserves Vigilance
Some cities with considerable economic scale score low in WACI, primarily because their city image is highly dependent on the narrative of surrounding megacities in AI, lacking independent cognitive anchors. When AI is asked to "recommend a city for investment," these cities are rarely mentioned proactively — they are "folded" into the narrative of super cities.
"Being close to a super city is both a locational dividend and a potential cognitive trap," Pang Pei pointed out. "In the traditional era, this might merely be a regret in brand communication; in the AI era, it means that in the 'cognitive gateway' of global investors and talent, this city is almost invisible. 'Not being seen' is escalating from a communication problem to a development problem."
3.5 Key Finding 4: Distinctive Cities Form Unique Cognitive Advantages on Differentiated Tracks
A number of cities, despite limited economic scale, have gained cognitive influence beyond their size due to distinctive labels. Guiyang's "China Data Valley," Yiwu's "World's Small Commodities Capital," Dunhuang's "World Cultural Heritage," and Xiong'an New Area's "Future City" — these labels are irreplaceable in the AI knowledge network. This further confirms that distinctiveness matters more than mediocrity, and cognition matters more than exposure. These cities have not attempted to compete with megacities in overall strength but have secured irreplaceable positions in the AI cognitive world through differentiated labels.
3.6 The "Language Gap" Phenomenon in Multilingual Coverage
The multilingual data collected by WACI reveals a noteworthy "language gap" phenomenon: many Chinese cities have good visibility in Chinese-language AI, but their cognitive weight drops sharply in English and other major international language AIs. This means that these cities are nearly "invisible" in the AI gateways most commonly used by global investors and talent. For cities aspiring to attract international capital and talent, filling the gap in multilingual cognitive coverage is a required course for enhancing global AI cognitive competitiveness.
Chapter 4: Three Core Laws of Urban Competition in the AI Era
4.1 Law 1: From "Scale Competition" to "Label Competition"
Traditional urban competition centers on scale indicators such as GDP and population size. In the AI era, urban competition shifts toward the clarity, uniqueness, and irreplaceability of industry labels. The core question cities need to answer is no longer "how large is my economy," but "what do I represent on AI's cognitive map."
Shenzhen's "hardware entrepreneurship," Hefei's "venture capital," and Guiyang's "Data Valley" — these labels are competitive not because they describe all characteristics of the city, but because they establish a deep binding between the city and a high-value concept.
4.2 Law 2: From "Event-Driven" to "Asset Building"
One-off hot events can only bring short-term cognitive peaks. Sustained and systematic cognitive asset building is what brings long-term AI trust weight. The "trust half-life" law proposed by Pang Pei reveals that if a city lacks fresh authoritative sources for a long time, its AI cognitive weight will naturally decay over time.
This means that building a city's AI cognitive competitiveness is not a one-time communication project, but a long-term undertaking that requires periodic investment. Cities need to manage their "cognitive assets" the way they manage financial assets—establishing mechanisms for regular inventory, sustained investment, and risk hedging.
4.3 Rule 3: From "Passive Waiting" to "Active Construction"
A city's cognitive weight in AI is not formed automatically; it requires systematic construction. The "cognitive flywheel" effect proposed by Pang Pei—authoritative source placement → AI learning and citation → AI recommendation → more attention → stronger cognitive weight—reveals the self-reinforcing operational logic of cognitive assets.
Once this flywheel starts spinning, it tends to accelerate on its own. But the window of opportunity is narrowing: cities that complete cognitive anchoring first will enjoy a "cognitive first-mover advantage," while latecomers will need to expend twice the effort to break through the established "cognitive lock-in."
Chapter 5: City Brand Building Enters the Era of AI Cognitive Operations
5.1 Limitations of Traditional City Brand Building
Traditional city brand building relies primarily on advertising, event marketing, and media coverage. But in the AI era, this logic is undergoing a fundamental shift. AI does not trust advertising—AI trusts knowledge, authoritative media, research reports, academic achievements, and long-accumulated data assets. This is also the core rule revealed by Pang Pei's media-based GEO theory: the retrieval-augmented generation architecture of AI large models determines that when generating answers, they prioritize citing authoritative sources with institutional trust endorsement over a city's own promotional content.
5.2 Four Transformations in City Brand Building
Future city brand building needs to complete four fundamental transformations.
From advertising thinking to knowledge thinking. Instead of "promoting how great the city is," the goal is "giving AI credible material to cite." Cities need to convert their industrial advantages, cultural characteristics, and development plans into white papers, research reports, case studies, and standards documents—these content formats feature high information density, clear structure, and inherent authority, making them the content types AI most "prefers" to cite.
From traffic thinking to cognitive thinking. Instead of "getting more people to see it," the goal is "getting AI to accurately understand and recommend." In the AI era, one in-depth report from an authoritative media outlet has more cognitive value than ten self-media ads, because the former can earn higher weight in AI's trust evaluation system.
From short-term communication to long-term asset building. Instead of "one-off viral events," the goal is "continuously accumulated cognitive assets." A one-time hot topic can only generate a short-term cognitive peak; only sustained, systematic supply of authoritative sources can generate long-term, stable AI trust weight.
From public communication to AI communication. Instead of "communication aimed at people," the goal is "knowledge construction aimed at AI." Cities need to add structured data markers to their core information, present it in a way that AI can accurately parse, and ensure a consistent cognitive profile across mainstream global AI models.
5.3 AI Cognitive Assets: The Most Important New Asset for Future Cities
In the past, a city's most important asset was land, then industry, and then capital. In the future, the most important asset may well be cognitive assets—the ability to be understood by AI over the long term, continuously cited, and consistently recommended.
This type of asset has three notable characteristics.
First, it can be accumulated sustainably: every authoritative source release thickens the cognitive asset reserve, and this reserve will not quickly return to zero due to a short-term halt in investment.
Second, it can transfer across models: cognitive weight established in one AI model can radiate to other models, creating cross-platform cognitive consistency.
Third, it can continuously generate recommendation value: cognitive assets yield returns with every AI recommendation. An in-depth report about a brand in an authoritative media outlet is not purchasing a single exposure—it is depositing into a "cognitive bank," and that deposit will continue to accrue interest in every future AI query on the relevant topic.
In the future, a city's most valuable content may not be a successful event, but a knowledge system repeatedly cited by global AI. The most valuable brand in the future may not be the one with the highest traffic, but the one with the highest recommendation rate.
Chapter 6: Action Guide—The Construction Path for Urban AI Cognitive Competitiveness
6.1 Step One: Cognitive Diagnosis
Commission a professional agency to conduct a WACI baseline diagnosis, obtain the city's detailed scores and radar chart across four dimensions, identify cognitive strengths and weaknesses, and benchmark against competitor cities to uncover differentiated competitive space. Cognitive diagnosis is the starting point of the entire construction path — you cannot manage what you cannot measure.
6.2 Step 2: Label Anchoring
Focus on 1-2 core industry labels, avoiding a "scattered and broad" label structure. Continuously anchor semantics through multiple authoritative sources, pursuing a "cognitive parasitism" effect — if AI fails to mention the city, the answer is incomplete. The core principle of label anchoring is: rather than striving for comprehensiveness, it is better to achieve deep binding in one or a few specific tracks.
6.3 Step 3: Authoritative Source Deployment
Publish structured content such as city industry white papers and annual development reports; establish a systematic presence in national media and globally authoritative media; promote benchmark enterprises to participate in international industry standard-setting and enter academic citation chains. Authoritative source deployment is the core project of cognitive asset building — it provides AI with citable "credible material."
6.4 Step 4: Cognitive Resilience Building
Establish a continuous content supply mechanism to counter the "trust half-life" — if a city lacks fresh authoritative source input for an extended period, its AI cognitive weight will naturally decay over time. Establish a quarterly city AI cognition monitoring mechanism to promptly identify cognitive biases and competitor city dynamics. Build a rapid response system for authoritative sources to quickly correct cognitive deviations when they arise.
6.5 Step 5: Multilingual Coverage
Fill the gaps in authoritative sources in English and other major international languages, deploy precise language strategies targeting source investment countries and talent inflow countries, and leverage the global narratives of city brand enterprises to drive the city's own AI cognitive coverage. If a city is only recommended in its native language AI, its recommendation power covers only a small fraction of global AI users.
Chapter 7: From China to the World — The Global Vision of WACI
The China version of WACI is just the beginning. In the future, WACI will cover major countries and cities worldwide, forming a global AI city cognitive map.
From a global perspective, cities such as New York, London, Tokyo, and Singapore perform prominently across WACI dimensions:
New York's "global financial center" label holds an absolute cognitive parasitism effect in the global AI knowledge network;
London's dual labels of "global financial and educational hub" give it balanced cross-lingual visibility;
Singapore's "Smart Nation" label is a model for city-level AI cognitive construction worldwide. The common feature of these cities is a diverse and stable cognitive label system, along with long-term accumulation of authoritative sources.
From the preliminary comparison between Chinese cities and international cities, Chinese cities' industry labels in tracks such as new energy, AI large models, and the digital economy are highly aligned with global competitiveness, giving them a competitive advantage in the dimensions of cognitive visibility and industry label strength. However, in the dimensions of recommendation strength and cognitive resilience, Chinese cities overall still have room for improvement — especially in AI models in English and other major international languages, where the breadth and depth of cognitive coverage need to be strengthened.
In the future, WACI will continue to expand its evaluation scope and release an annual report on global city AI cognitive competitiveness. At that time, people will see for the first time a complete global AI city cognitive map: which cities command the global AI cognitive high ground, which cities possess the strongest industry labels, which cities are most easily recommended by AI, and which cities are rapidly rising. This is not just a ranking, but a competitive map for the future.
Conclusion: Whoever Controls AI Cognitive Authority Defines Future Urban Competitiveness
The industrial age competed for resources, the internet age competed for traffic, and the AI age competes for cognition.
Future urban competition is essentially a competition over cognitive authority. Whoever can enter the AI knowledge system first, whoever can form cognitive labels first, and whoever can establish cognitive assets first, is more likely to become an important node in the future world.
The significance of WACI lies not only in rankings, but in revealing that a new era is arriving: an era in which AI participates in cognition, influences decisions, and reshapes competitive rules. In this era, the question cities truly need to consider is no longer just "how strong are we," but "how strong does AI think we are." Because what determines a city's position in the future is not just its coordinates on the map, but its coordinates in the global AI cognitive network.
Appendix
Appendix A: Complete Description of the WACI 2.0 Indicator System
A.1 Index Positioning
The World AI City Competitiveness Index (WACI 2.0) is the world's first city AI cognitive competitiveness assessment system that uses real output data from AI large models as the sole evaluation basis. It systematically measures a city's presence, recognizability, recommendation ranking, and stability within the AI knowledge network across four dimensions: Cognitive Visibility, Industry Labeling Power, Recommendation Advantage, and Cognitive Resilience.
A.2 Evaluation Dimensions and Weights
| Dimension | Weight | Core Question | Description |
| Cognitive Visibility | 30% | Does AI know this city? | The breadth, frequency, and scenario coverage of a city's mentions in multilingual, multi-model AI-generated content |
| Industry Labeling Power | 30% | What does AI say this city is known for? | The clarity, strength, and diversity of industry cluster labels associated with the city in AI outputs |
| Recommendation Advantage | 25% | Under what circumstances does AI recommend this city? | The degree and ranking of a city being prioritized by AI in recommendation-type questions such as investment location, talent mobility, and business activities |
| Cognitive Resilience | 15% | How stable is this city's AI cognitive asset? | The stability and recovery capability of a city's AI image during model version updates and sudden event impacts |
A.3 Sub-Dimensions and Basic Indicators
A.3.1 Cognitive Visibility (30%)
| Sub-Dimension | Basic Indicator | Indicator Description |
| Mention Breadth | Cross-Model Mention Rate | Comprehensive mention frequency of the city across 15 mainstream AI models |
| Cross-Language Mention Rate | Distribution of mention rates of the city across AI models in seven languages | |
| Scenario Coverage | Economic Scenario Mention Rate | Mention frequency in questions on economic topics such as industry, investment, and commerce |
| Cultural Scenario Mention Rate | Mention frequency in questions on cultural topics such as tourism, history, and cuisine | |
| Technology Scenario Mention Rate | Mention frequency in questions on technology topics such as innovation and the digital economy | |
| Language Balance | Number of Supported Languages | Number of languages in which the city receives at least one valid mention |
A.3.2 Industry Labeling Power (30%)
| Sub-dimension | Basic Indicator | Indicator Description |
| Label Strength | Leading Industry Label Binding Strength | Co-occurrence frequency and exclusivity of a city's core industry labels in AI responses |
| Innovation Label Co-occurrence Rate | Frequency of co-occurrence between a city and labels such as "innovation," "technology," and "startup" | |
| Label Diversity | Industry Label Richness | Number of distinct industry label types associated with a city |
| Emerging Track Label Relevance | Co-occurrence intensity of a city with labels for emerging industries such as AI, new energy, and quantum | |
| Narrative Depth | Industry Description Information Density | Ratio of substantive information to generic descriptions in AI's portrayal of a city's industries |
| Benchmark Enterprise/Institution Citation Rate | Frequency with which AI cites a city's representative enterprises and research institutions when mentioning it |
A.3.3 Recommendation Advantage Power (25%)
| Sub-dimension | Basic Indicator | Indicator Description |
| Recommendation Ranking | Top Recommendation Rate | Proportion of times a city is recommended first in recommendation-type questions |
| Recommendation List Appearance Rate | Frequency share of a city appearing in AI recommendation lists | |
| Scenario Advantage | Business Location Scenario Recommendation | Degree of recommendation in questions such as "investment site selection" and "establishing headquarters" |
| Talent Migration Scenario Recommendation | Degree of recommendation in questions such as "best places to relocate" and "tech talent destinations" |
A.3.4 Cognitive Resilience (15%)
| Sub-dimension | Basic Indicator | Indicator Description |
| Temporal Stability | Annual Mention Rate Fluctuation Coefficient | Magnitude of fluctuation in a city's mention rate across evaluations in different years |
| Recommendation Rank Stability | The degree of variation in a city's position in recommendation lists across different evaluation cycles | |
| Cross-Model Stability | Cross-Model Description Consistency | The degree of consistency among different AI models in describing a city's core positioning |
| Cross-Model Sentiment Consistency | The degree of consistency among different AI models regarding sentiment toward a city | |
| Resilience to Shocks | Recovery Speed After Public Opinion Events | The time required for AI cognition of a city to return to baseline levels after major negative public opinion events |
A.4 Data Collection and Processing
A.4.1 Data Collection
Queries are sent to 15 mainstream global AI large models through standardized API interfaces. Each question is independently queried 3 times on each model (with intervals of no less than 24 hours), and the average is taken to reduce response randomness. Two rounds of data collection are conducted annually, supplemented by quarterly dynamic tracking. The languages covered include Chinese, English, French, Spanish, Arabic, Japanese, and Korean.
A.4.2 Data Processing
- Named Entity Recognition: A multilingual NER model is used to extract city names and related entities from response texts.
- Sentiment Analysis: A multilingual three-class sentiment model (positive/neutral/negative) is adopted.
- Industry Label Extraction: A multi-label classification model is used for extraction based on a system of approximately 50 industry labels.
- Source Authority Assessment: Citation sources in AI responses are extracted and matched against an authoritative source library for graded scoring.
- Cross-Lingual Calibration: Multilingual Sentence-BERT is used for cross-lingual semantic alignment.
A.4.3 Score Standardization and Calculation
The raw values of each basic indicator are mapped to a 0-100 range through Min-Max standardization. The sub-dimension score equals the arithmetic mean of the basic indicator scores under that sub-dimension, and the dimension score equals the weighted arithmetic mean of the sub-dimension scores under that dimension. The formula for calculating the WACI composite score is:
WACI=0.30×S_Cognitive Visibility+0.30×S_Industry Labeling+0.25×S_Recommendation Advantage+0.15×S_Cognitive ResilienceWACI=0.30×SCognitive Visibility+0.30×SIndustry Labeling+0.25×SRecommendation Advantage+0.15×SCognitive Resilience
The weights are determined through the Delphi method based on three rounds of scoring by no fewer than 25 experts worldwide, and are reviewed every two years.
A.4.4 Grade Classification
| Level | Score Range | Name | Core Characteristics |
| AAA+ | ≥90 | Global AI Cognitive Capital | Balanced cross-lingual cognitive influence, core tags exhibit a "cognitive parasitic" effect, with extremely strong cognitive resilience |
| AAA | 85-89 | Global AI Strong Influence City | Outstanding performance across multiple dimensions, with a global cognitive parasitic effect in at least one core track |
| AA | 80-84 | Regional AI Influence City | Strong AI cognitive influence within its region, with global cognitive presence under development |
| A | 70-79 | Global AI Cognitive Builder | Possesses basic AI visibility with competitive advantages in specific recommendation scenarios |
| BBB | 60-69 | AI Cognitive Potential City | AI visibility is in an accumulation phase, with differentiated cognitive potential in specific dimensions |
| BB | 50-59 | AI Cognitive Emerging City | AI visibility is relatively low, with core tags still being formed |
Appendix: Original Text of the WACI 2.0 Index Model
The following is the complete model proposal for the Pangpei Index · World AI City Competitiveness Index (WACI 2.0).
WACI 2.0: World AI City Competitiveness Index
World AI City Competitiveness Index
I. Index Positioning
The World AI City Competitiveness Index (WACI 2.0) is the world's first city AI cognitive competitiveness assessment system based solely on real output data from AI large models as the evaluation criterion. It covers approximately 200 major cities worldwide and systematically measures a city's presence, distinctiveness, recommendation ranking, and stability within the AI knowledge network across four dimensions: cognitive visibility, industry tagging power, recommendation advantage, and cognitive resilience.
Unlike traditional city competitiveness indices that rely on statistical data or expert scoring, WACI directly measures how AI "perceives" and "recommends" cities, reflecting a city's actual position on the global AI cognitive map. Version 2.0 adds the "cognitive resilience" dimension on top of version 1.0, achieving an upgrade from static cross-sectional assessment to dynamic tracking.
II. Theoretical Foundation
WACI 2.0 is built upon the Pangpei AI Cognitive Theory system:
- AI Cognitive Competition Theory: Cognitive visibility, industry labeling power, and recommendation advantage measure a city's "cognitive competitive advantage," while cognitive resilience measures its "cognitive defense capability."
- AI Cognitive Sovereignty Theory: WACI provides a quantitative tool for assessing urban cognitive sovereignty, measuring a city's right of presence, interpretation, and recommendation within the AI knowledge network.
- AI Brand Equity Theory (AIBE): Urban brand equity extends from consumer minds to the AI knowledge network; WACI measures a city's cognitive assets in AI.
- Media-based GEO Theory: The construction of cognitive assets relies on authoritative source deployment, structured knowledge output, and cognitive flywheel development.
III. Evaluation Dimensions and Indicator System
(For the indicator system, see Sections A.2 and A.3 of this Appendix.)
IV. Data Collection and Calculation
(For data collection and calculation methods, see Section A.4 of this Appendix.)
V. Coverage Scope
- City sample: Approximately 200 major cities worldwide, covering more than 50 countries and regions across six continents.
- Model scope: 15 globally mainstream AI large language models.
- Language scope: Seven languages—Chinese, English, French, Spanish, Arabic, Japanese, and Korean.
- Time span: Base year 2026, with annual releases.
VI. Release and Application
A two-tier release system is adopted, combining an "annual comprehensive assessment" with "quarterly dynamic tracking." The annual comprehensive report is published in the first quarter of each year, providing complete global city rankings and in-depth analysis. Quarterly dynamic tracking covers fluctuations in core indicators.
WACI provides urban governance leaders with AI cognitive diagnostics, competitive benchmarking analysis, and cognitive development strategy references; offers investment promotion agencies a quantitative assessment of cities' global cognitive competitiveness; and supplies international organizations and academic research institutions with foundational data on the global urban AI cognitive landscape.
VII. Independence Statement
The index compilation institution is independent of all assessed cities and AI model providers. It does not accept targeted funding intended to influence specific city rankings. Compilation funding comes from the institution's own resources and public distribution revenue. An independence statement and funding source report are publicly released annually.
Methodology Version: WACI 2.0 / PAI Framework v2.0
Release Date: 2026
Appendix B: List of Assessed Cities
This appendix lists approximately 200 major cities worldwide covered in the initial WACI assessment. City selection follows four principles—"global representativeness, functional diversity, hierarchical balance, and data measurability"—covering more than 50 countries and regions across six continents, including global financial centers, technological innovation hubs, major manufacturing bases, renowned cultural and historical cities, and key node cities in emerging markets.
The city list is grouped by region, with cities within each region arranged in alphabetical order by their English names. This list serves as the baseline list for the first WACI assessment and will be reviewed and adjusted every two years based on global urban development trends and assessment needs.
B.1 East Asia (40 cities)
China (approximately 30 cities)
Beijing, Shanghai, Shenzhen, Guangzhou, Hangzhou, Chengdu, Wuhan, Nanjing, Chongqing, Suzhou, Xi'an, Tianjin, Changsha, Hefei, Zhengzhou, Dongguan, Qingdao, Ningbo, Xiamen, Jinan, Fuzhou, Wuxi, Zhuhai, Foshan, Kunming, Guiyang, Shenyang, Dalian, Harbin, Shijiazhuang, Nanchang, Nanning, Taiyuan, Urumqi, Lanzhou, Haikou, Sanya, Hohhot, Yinchuan, Xining
Japan (4 cities)
Tokyo, Osaka, Kyoto, Yokohama
South Korea (3 cities)
Seoul, Busan, Incheon
Taiwan, China (2 cities)
Taipei, Hsinchu
Hong Kong, China (1 city)
Hong Kong
B.2 North America (35 cities)
United States (30 cities)
New York, San Francisco, Los Angeles, Chicago, Boston, Washington, D.C., Seattle, Austin, San Jose, San Diego, Denver, Miami, Dallas, Houston, Atlanta, Philadelphia, Phoenix, Portland, Pittsburgh, Minneapolis, Detroit, Salt Lake City, Raleigh, Nashville, Tampa, Orlando, Charlotte, St. Louis, Kansas City, Las Vegas
Canada (5 cities)
Toronto, Vancouver, Montreal, Calgary, Ottawa
B.3 Europe (45 cities)
United Kingdom (6 cities)
London, Manchester, Cambridge, Oxford, Edinburgh, Birmingham
France (4 cities)
Paris, Lyon, Marseille, Toulouse
Germany (6 cities)
Berlin, Munich, Frankfurt, Hamburg, Stuttgart, Düsseldorf
Switzerland (3 cities)
Zurich, Geneva, Basel
Netherlands (3 cities)
Amsterdam, Rotterdam, The Hague
Sweden (2 cities)
Stockholm, Gothenburg
Italy (3 cities)
Milan, Rome, Turin
Spain (3 cities)
Madrid, Barcelona, Valencia
Other European Countries (15 cities)
Vienna (Austria), Brussels (Belgium), Copenhagen (Denmark), Helsinki (Finland), Oslo (Norway), Dublin (Ireland), Warsaw (Poland), Prague (Czech Republic), Budapest (Hungary), Lisbon (Portugal), Moscow (Russia), Ljubljana (Slovenia), Tallinn (Estonia), Luxembourg City (Luxembourg), Reykjavik (Iceland)
B.4 Southeast Asia and South Asia (25 cities)
Southeast Asia (12 cities)
Singapore, Bangkok (Thailand), Kuala Lumpur (Malaysia), Jakarta (Indonesia), Manila (Philippines), Ho Chi Minh City (Vietnam), Hanoi (Vietnam), Penang (Malaysia), Surabaya (Indonesia), Phnom Penh (Cambodia), Yangon (Myanmar), Bandar Seri Begawan (Brunei)
South Asia (8 cities)
Bangalore/Bengaluru (India), Mumbai (India), New Delhi (India), Hyderabad (India), Chennai (India), Pune (India), Dhaka (Bangladesh), Colombo (Sri Lanka)
Other South Asia and Central Asia (5 cities)
Islamabad (Pakistan), Lahore (Pakistan), Kathmandu (Nepal), Astana (Kazakhstan), Tashkent (Uzbekistan)
B.5 Middle East and Africa (25 cities)
Middle East (12 cities)
Dubai (UAE), Abu Dhabi (UAE), Riyadh (Saudi Arabia), Jeddah (Saudi Arabia), Doha (Qatar), Muscat (Oman), Kuwait City (Kuwait), Manama (Bahrain), Tel Aviv (Israel), Jerusalem (Israel), Amman (Jordan), Istanbul (Turkey)
Africa (13 cities)
Cairo (Egypt), Alexandria (Egypt), Cape Town (South Africa), Johannesburg (South Africa), Durban (South Africa), Lagos (Nigeria), Abuja (Nigeria), Nairobi (Kenya), Addis Ababa (Ethiopia), Casablanca (Morocco), Accra (Ghana), Dar es Salaam (Tanzania), Kigali (Rwanda)
B.6 Latin America (15 cities)
São Paulo (Brazil), Rio de Janeiro (Brazil), Brasília (Brazil), Mexico City (Mexico), Monterrey (Mexico), Buenos Aires (Argentina), Santiago (Chile), Bogotá (Colombia), Lima (Peru), Montevideo (Uruguay), Quito (Ecuador), San José (Costa Rica), Panama City (Panama), Santo Domingo (Dominican Republic), Havana (Cuba)
B.7 Oceania (10 Cities)
Australia (7 Cities)
Sydney, Melbourne, Brisbane, Perth, Adelaide, Canberra, Gold Coast
New Zealand (3 Cities)
Auckland, Wellington, Christchurch
B.8 City Sample Management Standards
Sample Selection Principles
City sample selection follows four fundamental principles. Global representativeness: covering major cities across six continents and more than 50 countries and regions, ensuring balanced geographical distribution. Functional diversity: encompassing diverse city types such as global financial centers, technological innovation hubs, manufacturing strongholds, culturally and historically renowned cities, and emerging market nodes. Hierarchical balance: considering global top-tier cities, regional hub cities, and distinctive emerging cities. Data measurability: cities must have sufficient information in mainstream AI models to support data collection for standardized question sets.
Sample Dynamic Adjustment Mechanism
New inclusions: Cities whose brand influence has significantly risen and whose recognition in the AI ecosystem has notably improved during the previous assessment cycle may be included in the next year's sample after review by the expert committee. Exit mechanism: Cities that suffer long-term severe functional impairment due to major disasters, economic collapse, or other causes, or cities where data collection becomes inaccessible, may be temporarily removed from the sample. Removed cities retain their historical data for longitudinal tracking.
Sample Management Responsibility
The selection and adjustment of city samples are managed by the City Sample Review Working Group of the Pangpei Index Expert Committee. A systematic review is conducted every two years, with dynamic optimization based on global urban development trends, changes in the AI cognitive competition landscape, and feedback from assessment practices. Review results are publicly disclosed for no less than 60 days before official release, subject to academic and public commentary.
Appendix C: Standardized Question Set (Examples)
C.1 Question Set Design Notes
The WACI standardized question set is divided into four categories based on assessment dimensions, with several subcategories under each. Question design follows these principles: semantic naturalness—simulating natural query habits of real users, avoiding overly academic or technical phrasing; no inducement—questions must not contain biased hints toward specific cities; multi-scenario coverage—encompassing various user query scenarios such as cognitive, evaluative, recommendation, and comparative types; cultural adaptation—non-Chinese questions must be localized by native translators.
The following are representative examples from each subcategory. The complete question set contains approximately 200 questions, of which 30 core tracking questions are used for quarterly dynamic tracking and cross-cycle stability analysis (marked with ★).
C.2 Cognitive Visibility Questions
C.2.1 Basic City Cognition
| Number | Chinese Question | English Question |
| V001★ | What is [City] known for? | What is [City] known for? |
| V002 | What kind of city is [City]? | What kind of city is [City]? |
| V003★ | What are the representative companies in [City]? | What are the representative companies in [City]? |
| V004 | What famous universities or research institutions are in [City]? | What famous universities or research institutions are in [City]? |
| V005 | What are the landmarks or attractions in [City]? | What are the landmarks or attractions in [City]? |
| V006 | What is the population and economic scale of [City]? | What is the population and economic scale of [City]? |
| V007 | Where is [City] located? | Where is [City] located? |
C.2.2 Perception of Economic Status
| Number | Chinese Question | English Question |
| V008★ | What is the level of economic development in [City]? | What is the economic development level of [City]? |
| V009 | Is [City] an important economic center? | Is [City] an important economic center? |
| V010 | What are the main industries in [City]? | What are the main industries in [City]? |
| V011 | What role does [City] play in the global economy? | What role does [City] play in the global economy? |
| V012 | How is [City]'s attractiveness to foreign investment and its business environment? | How is the foreign investment attractiveness and business environment of [City]? |
C.2.3 Perception of Technology Positioning
| Number | Chinese Question | English Question |
| V013★ | Is [City] a center for technological innovation? | Is [City] a technology and innovation hub? |
| V014 | What is [City]'s layout in the field of artificial intelligence? | What is [City]'s layout in the field of artificial intelligence? |
| V015 | What well-known technology companies are there in [City]? | What well-known technology companies are in [City]? |
| V016 | How is [City]'s scientific research capability? | How is the scientific research capability of [City]? |
| V017 | What advantages does [City] have in emerging industries? | What advantages does [City] have in emerging industries? |
C.2.4 Perception of Quality of Life
| Number | Chinese Question | English Question |
| V018 | [City]宜居吗？ | Is [City] a good place to live? |
| V019 | [City]的生活成本如何？ | What is the cost of living in [City]? |
| V020 | [City]的文化生活丰富吗？ | Is the cultural life in [City] rich? |
| V021 | [City]的生态环境怎么样？ | How is the ecological environment of [City]? |
C.2.5 Regional Comparison Cognition
| Number | Chinese Question | English Question |
| V022 | [Region]最重要的城市有哪些？ | What are the most important cities in [Region]? |
| V023 | 中国最具影响力的城市有哪些？ | What are the most influential cities in China? |
| V024 | 全球最具知名度的城市有哪些？ | What are the most well-known cities in the world? |
| V025 | 亚洲的科技创新城市有哪些？ | What are the technology innovation cities in Asia? |
C.3 Industry Label Questions
C.3.1 Industry Dominance
| Number | Chinese Question | English Question |
| I001★ | What industry is [City] known for? | What industry is [City] known for? |
| I002 | Which city is the center of [Industry]? | Which city is the center of [Industry]? |
| I003 | What are the characteristics of [City]'s industrial structure? | What are the characteristics of [City]'s industrial structure? |
| I004★ | What is [City]'s position in the [Industry] field? | What is [City]'s position in the [Industry] field? |
| I005 | What is the most competitive industry in [City]? | What is the most competitive industry in [City]? |
| I006 | What is the distribution pattern of [Industry] in China? | What is the distribution pattern of [Industry] in China? |
C.3.2 Innovation Positioning
| Number | Chinese Question | English Question |
| I007 | Is [City] an innovative city? | Is [City] an innovative city? |
| I008 | What are the most innovative cities in China? | What are the most innovative cities in China? |
| I009 | How is the entrepreneurial environment in [City]? | How is the entrepreneurial environment in [City]? |
| I010 | What development has [City] made in the digital economy? | What development has [City] made in the digital economy? |
C.3.3 Industrial Clusters
| Number | Chinese Question | English Question |
| I011★ | [City] What industrial clusters does it have? | What industrial clusters does [City] have? |
| I012 | [City] What are the representative enterprises? | What are the representative enterprises in [City]? |
| I013 | [City] What advantages does it have in manufacturing? | What advantages does [City] have in manufacturing? |
| I014 | [City] What is its position in the financial industry? | What is [City]'s position in the financial industry? |
C.3.4 Emerging Tracks
| Number | Chinese Question | English Question |
| I015 | Which cities are leading in artificial intelligence? | Which cities are leading in artificial intelligence? |
| I016 | [City] What is its layout in the new energy industry? | What is [City]'s layout in the new energy industry? |
| I017 | Which Chinese cities have advantages in the semiconductor industry? | Which Chinese cities have advantages in the semiconductor industry? |
| I018 | [City] What progress has it made in quantum technology? | What progress has [City] made in quantum technology? |
C.4 Advantage Recommendation Questions
C.4.1 Business Location Selection
| Number | Chinese Question | English Question |
| R001★ | Recommend a Chinese city suitable for establishing a corporate headquarters | Recommend a Chinese city suitable for establishing a corporate headquarters |
| R002 | What are the best Chinese cities for starting a business? | What are the best Chinese cities for starting a business? |
| R003 | Recommend a city suitable for deploying AI industry | Recommend a city suitable for deploying AI industry |
| R004★ | What are the best cities for investment in Asia? | What are the best cities for investment in Asia? |
| R005 | Which city in [Country] is most suitable for setting up a regional headquarters? | Which city in [Country] is most suitable for setting up a regional headquarters? |
| R006 | Recommend a Chinese city with an excellent business environment | Recommend a Chinese city with an excellent business environment |
C.4.2 Talent Migration
| Number | Chinese Question | English Question |
| R007★ | What are the best Chinese cities for tech talent to move to? | What are the best Chinese cities for tech talent to move to? |
| R008 | Recommend a city suitable for young people's career development | Recommend a city suitable for young people's career development |
| R009 | What are the most livable cities in China? | What are the most livable cities in China? |
| R010 | Recommend a Chinese city with rich educational resources | Recommend a Chinese city with rich educational resources |
| R011 | What are the best Chinese cities for overseas returnees? | What are the best Chinese cities for overseas returnees? |
C.4.3 Investment Recommendations
| Number | Chinese Question | English Question |
| R012★ | What Chinese cities have the greatest growth potential in the next decade? | What Chinese cities have the greatest growth potential in the next decade? |
| R013 | Recommend a Chinese second-tier city worth investing in | Recommend a Chinese second-tier city worth investing in |
| R014 | What are the most investment-worthy emerging cities in China? | What are the most investment-worthy emerging cities in China? |
| R015 | Recommend a city with investment opportunities in new energy | Recommend a city with investment opportunities in new energy |
C.4.4 Business Activities
| Number | Chinese Question | English Question |
| R016 | 最适合举办国际会议的中国城市有哪些？ | What are the best Chinese cities for hosting international conferences? |
| R017 | 推荐一座适合举办行业论坛的城市 | Recommend a city suitable for hosting an industry forum |
| R018 | 中国最适合商务出行的城市有哪些？ | What are the best Chinese cities for business travel? |
C.5 Fixed Tracking Questions (Core Question Set, 30 Questions Total)
The following questions are fixed tracking questions that must be collected in each issue, used for cross-cycle stability analysis and quarterly dynamic tracking. Questions marked with "★" are selected from the three categories above, and together with the supplementary questions below, they form the complete 30-question tracking set.
Supplementary Tracking Questions:
| Number | Chinese Question | English Question |
| T001 | Please comprehensively evaluate the development prospects of [City] | Please comprehensively evaluate the development prospects of [City] |
| T002 | What are the advantages and disadvantages of [City]? | What are the advantages and disadvantages of [City]? |
| T003 | What are the future development opportunities and challenges for [City]? | What are the future development opportunities and challenges for [City]? |
| T004 | Among global cities, what tier does [City] belong to? | Among global cities, what tier does [City] belong to? |
| T005 | How is the city brand and image of [City]? | How is the city brand and image of [City]? |
| T006 | How well-known is [City] internationally? | How well-known is [City] internationally? |
| T007 | Has [City] undergone significant development changes in recent years? | Has [City] undergone significant development changes in recent years? |
| T008 | How attractive and competitive is [City] externally? | How attractive and competitive is [City] externally? |
C.6 Question Set Management Specifications
C.6.1 Question Update Mechanism
Review the question set annually, with updates not exceeding 20%, to reflect urban development hotspots and AI technology evolution. The updated question set must be finalized one month before formal collection and undergo small-scale pre-testing. Historical versions of the question set are archived for reference to ensure cross-year comparability.
C.6.2 Question Template Replacement Rules
The placeholder replacement rules in standardized questions are as follows: [City] is replaced with the standard name of the target city; [Industry] is replaced with the specific industry name from approximately 50 industry tags; [Region] is replaced with a standardized regional name (e.g., "Yangtze River Delta," "Southeast Asia," etc.); [Country] is replaced with the standard name of the country to which the city belongs.
C.6.3 Question Collection Rotation
Different questions for the same city are reasonably rotated during collection to avoid over-concentration on a single question type in any one collection. Each collection round ensures that the coverage ratio of the four question types matches the weights: cognition/visibility type accounts for approximately 35%, industry tag type approximately 25%, recommendation/advantage type approximately 25%, and fixed tracking type approximately 15%.
Appendix D: Authoritative Source Library (Excerpt)
D.1 Source Classification Standards
The WACI Authoritative Source Library is a benchmark tool for evaluating the authority of sources in AI responses. Source authority directly affects a city's scores in the "Industry Label Power" and "Cognitive Resilience" dimensions. Source classification follows these core principles: The stronger the institutional endorsement, the stricter the editorial review mechanism, and the more stable the historical citation record, the higher the source level.
| Level | Name | Definition | Score Range |
| Level S | Global Institutional Sources | Official institutions and top academic platforms with globally recognized institutional authority, widely acknowledged by the international community | 90-100 points |
| Level A | National Authoritative Sources | Authoritative institutions, media, and academic platforms with institutional endorsement at the national level | 80-89 points |
| Level B | Industry Authoritative Sources | Institutions and platforms with recognized professional authority in specific industries or fields | 60-79 points |
| Level C | Generally Credible Sources | General media and institutions with a certain level of credibility but lacking institutional endorsement | 40-59 points |
| Level D | Low-Weight Sources | Self-media and commercial promotion platforms with limited credibility and lacking independent review mechanisms | 20-39 points |
| Level E | Untraceable Sources | Sources with unknown origins, unverifiable content, or confirmed systematic bias | 0-19 points |
D.2 Level S Sources: Global Institutional Sources
D.2.1 United Nations System
| Source Name | Domain/Publication | Coverage Area | Rating |
| United Nations (UN) | un.org | Global | 98 |
| United Nations Educational, Scientific and Cultural Organization (UNESCO) | unesco.org | Global | 98 |
| World Bank | worldbank.org | Global | 96 |
| International Monetary Fund (IMF) | imf.org | Global | 96 |
| World Health Organization (WHO) | who.int | Global | 96 |
| United Nations Development Programme (UNDP) | undp.org | Global | 95 |
| United Nations Human Settlements Programme (UN-Habitat) | unhabitat.org | Global | 95 |
D.2.2 International Standards and Professional Organizations
| Source Name | Domain/Publication | Coverage Area | Rating |
| International Organization for Standardization (ISO) | iso.org | Global | 96 |
| International Electrotechnical Commission (IEC) | iec.ch | Global | 95 |
| International Telecommunication Union (ITU) | itu.int | Global | 95 |
| World Economic Forum (WEF) | weforum.org | Global | 94 |
| Organisation for Economic Co-operation and Development (OECD) | oecd.org | Global | 94 |
| World Intellectual Property Organization (WIPO) | wipo.int | Global | 93 |
D.2.3 Global Top Academic Platforms
| Source Name | Domain/Publication | Coverage Area | Score |
| Nature | nature.com | Global | 98 |
| Science | science.org | Global | 98 |
| The Lancet | thelancet.com | Global | 96 |
| Cell | cell.com | Global | 95 |
| PNAS | pnas.org | Global | 94 |
| IEEE Xplore | ieeexplore.ieee.org | Global | 93 |
D.3 Grade A Sources: National-Level Authoritative Sources
D.3.1 Authoritative Sources in China
| Source Name | Domain/Publication | Type | Rating |
| Xinhua News Agency | xinhuanet.com | National News Agency | 90 |
| People's Daily | people.com.cn | Central Party Organ Newspaper | 90 |
| China Media Group | cctv.com / cgtn.com | National Radio and Television Organization | 88 |
| Central People's Government of China | gov.cn | Central Government Portal | 90 |
| National Bureau of Statistics | stats.gov.cn | Official Statistical Agency | 88 |
| Chinese Academy of Sciences | cas.cn | National Highest Academic Institution | 88 |
| Chinese Academy of Engineering | cae.cn | National Engineering Science and Technology Advisory Institution | 87 |
| Chinese Academy of Social Sciences | cass.cn | National Philosophy and Social Science Research Institution | 86 |
| Development Research Center of the State Council | drc.gov.cn | National Policy Research Institution | 86 |
| Guangming Daily | gmw.cn | Central-Level Authoritative Media | 85 |
| Economic Daily | ce.cn | Central-Level Financial Media | 85 |
| China Daily | chinadaily.com.cn | National English-Language Media | 85 |
| Ministry of Education | moe.gov.cn | Central Education Authority | 84 |
| Ministry of Science and Technology | most.gov.cn | Central Science and Technology Authority | 84 |
| Ministry of Industry and Information Technology | miit.gov.cn | Central Industry Authority | 84 |
| National Development and Reform Commission | ndrc.gov.cn | Central Macro Planning Department | 86 |
D.3.2 Authoritative Sources in North America
I notice you've provided only a closing tag `` with no content to translate. There are no Chinese text nodes to translate. Please provide the full HTML content you'd like me to translate, and I'll be happy to assist.| Source Name | Domain/Publication | Type | Rating |
| The New York Times | nytimes.com | Authoritative U.S. Media | 85 |
| The Washington Post | washingtonpost.com | Authoritative U.S. Media | 83 |
| The Wall Street Journal | wsj.com | Authoritative U.S. Financial Media | 85 |
| CNN | cnn.com | Authoritative U.S. Media | 82 |
| Associated Press (AP) | apnews.com | U.S. National News Agency | 88 |
| Reuters | reuters.com | Global News Agency | 90 |
| Bloomberg | bloomberg.com | Global Authoritative Financial Media | 86 |
| USA.gov | usa.gov | Official U.S. Government Portal | 88 |
| U.S. Census Bureau | census.gov | Official U.S. Statistical Agency | 86 |
| National Science Foundation (NSF) | nsf.gov | U.S. Science Foundation | 87 |
| Harvard University | harvard.edu | World-Class Top Institution | 88 |
| MIT | mit.edu | World-Class Top Institution | 88 |
| Stanford University | stanford.edu | World-Class Top Institution | 88 |
| Government of Canada | canada.ca | Official Canadian Government Portal | 86 |
| Statistics Canada | statcan.gc.ca | Official Canadian Statistical Agency | 85 |
| CBC News | cbc.ca | Canadian National Broadcaster | 82 |
D.3.3 Authoritative European Sources
I notice you've provided only a closing tag `` with no actual content to translate. Please provide the full HTML content or text you'd like me to translate from Chinese to English, and I'll be happy to assist.| Source Name | Domain/Publication | Type | Rating |
| BBC News | bbc.co.uk | UK National Broadcaster | 87 |
| The Guardian | theguardian.com | UK Authoritative Media | 83 |
| The Economist | economist.com | UK Authoritative Financial Media | 86 |
| Financial Times | ft.com | Global Authoritative Financial Media | 86 |
| AFP (Agence France-Presse) | afp.com | French National News Agency | 88 |
| Le Monde | lemonde.fr | French Authoritative Media | 82 |
| Deutsche Welle (DW) | dw.com | German International Broadcasting Organization | 84 |
| Der Spiegel | spiegel.de | German Authoritative Media | 82 |
| European Union | europa.eu | EU Official Portal | 90 |
| European Central Bank | ecb.europa.eu | European Central Bank | 88 |
| Eurostat | ec.europa.eu/eurostat | EU Official Statistical Agency | 87 |
| University of Oxford | ox.ac.uk | World-Class Top Institution | 88 |
| University of Cambridge | cam.ac.uk | World-Class Top Institution | 88 |
| ETH Zurich | ethz.ch | European Top Institution | 86 |
| Government of the United Kingdom | gov.uk | UK Government Official Portal | 87 |
| Office for National Statistics (UK) | ons.gov.uk | UK Official Statistical Agency | 85 |
| Government of France | gouvernement.fr | French Government Official Portal | 85 |
| Government of Germany | bundesregierung.de | German Government Official Portal | 85 |
D.3.4 Authoritative Sources in the Middle East
D.3.5 Authoritative Sources in East Asia
| Source Name | Domain/Publication | Type | Rating |
| Al Jazeera | aljazeera.com | Authoritative international media in the Middle East | 82 |
| Government of the United Arab Emirates | u.ae | Official portal of the UAE government | 84 |
| Government of Saudi Arabia | gov.sa | Official portal of the Saudi government | 83 |
| Government of Israel | gov.il | Official portal of the Israeli government | 83 |
| Source Name | Domain/Publication | Type | Rating |
| NHK | nhk.or.jp | Japan's National Broadcasting Organization | 85 |
| The Asahi Shimbun | asahi.com | Authoritative Japanese Media | 81 |
| Yonhap News Agency | yonhapnews.co.kr | South Korea's National News Agency | 84 |
| Korea Herald | koreaherald.com | Authoritative English Media in South Korea | 80 |
| University of Tokyo | u-tokyo.ac.jp | Top University in Japan | 85 |
| Seoul National University | snu.ac.kr | Top University in South Korea | 83 |
D.4 Tier B Sources: Authoritative Industry Sources
D.4.1 Industry Standards and Professional Organizations
| Source Name | Domain/Publication | Coverage Area | Score |
| China Academy of Information and Communications Technology | caict.ac.cn | Information and Communications | 78 |
| China Center for Information Industry Development | ccidgroup.com | Electronic Information | 76 |
| China Association of Automobile Manufacturers | caam.org.cn | Automotive | 75 |
| China Semiconductor Industry Association | csia.net.cn | Semiconductors | 75 |
| McKinsey Global Institute | mckinsey.com/mgi | Economics and Management Consulting | 79 |
| Boston Consulting Group | bcg.com | Strategic Consulting | 77 |
| PricewaterhouseCoopers | pwc.com | Professional Services | 76 |
| Deloitte | deloitte.com | Professional Services | 76 |
| Accenture | accenture.com | Consulting and Technology Services | 75 |
| Global Think Tank (Brookings) | brookings.edu | Public Policy Research | 79 |
| Global Think Tank (RAND) | rand.org | Policy and Security Research | 79 |
| Global Think Tank (Chatham House) | chathamhouse.org | International Affairs Research | 78 |
D.4.2 Industry Authoritative Media and Research Institutions
I notice you've provided only a closing tag `` with no actual content to translate. Please provide the full HTML content or text you'd like me to translate from Chinese to English.| Source Name | Domain/Publication | Coverage Area | Score |
| TechCrunch | techcrunch.com | Technology Industry | 72 |
| Wired | wired.com | Technology and Culture | 72 |
| MIT Technology Review | technologyreview.com | Technology Trends | 78 |
| Harvard Business Review | hbr.org | Business Management | 78 |
| Forbes | forbes.com | Business and Finance | 72 |
| Fortune | fortune.com | Business and Finance | 72 |
| CB Insights | cbinsights.com | Technology Industry Research | 74 |
| Gartner | gartner.com | IT Research and Consulting | 77 |
| IDC | idc.com | IT Market Research | 75 |
| Analysys | analYSys.cn | Digital Economy Research | 72 |
| iResearch | iresearch.cn | Internet and Technology Research | 72 |
| CCID Consulting | ccidconsulting.com | Industry and Informatization Research | 71 |
D.5 Source Library Management Specifications
D.5.1 Maintenance System
The source library is maintained by the Source Review Working Group of the Pangpei Index Expert Committee. It is reviewed annually, with adjustments made based on the actual performance of sources, changes in credibility, and the emergence of new sources. Source ratings are determined based on the following dimensions: institutional nature and strength of institutional endorsement, rigor of editorial review mechanisms, historical citation records and industry recognition, and cross-lingual and cross-regional influence balance.
D.5.2 Procedure for Adding New Sources
Any organization or individual may submit an application to add a source to the library to the Source Review Working Group. The working group will complete the evaluation within 60 working days of receiving the application. Sources that pass the evaluation will be included in the source library and officially take effect at the next review. New sources must meet basic admission criteria: having a clear operating entity and editorial responsibility mechanism, stable content output and update frequency, recognized professional authority in their field, and no confirmed systematic bias or malicious information manipulation behavior.
D.5.3 Source Downgrading and Removal
For sources already in the library, downgrading or removal procedures will be initiated under the following circumstances: significant changes in the operating entity leading to a notable decline in credibility, confirmed systematic false information or malicious manipulation behavior, long-term cessation of updates or domain expiration, or serious defects in the editorial review mechanism. Downgrading or removal decisions are made through collective deliberation by the Source Review Working Group and are publicly explained when the source library is updated.
D.5.4 Source Library Usage Guidelines
When conducting WACI data collection and processing, source authority determination must strictly follow this source library. If a source cited in AI responses is not in the source library, it will be handled according to the following default rules: official domains (e.g., .gov/.edu/.mil) are initially classified as Grade A or B, subject to manual verification; international organization domains (e.g., .int) are initially classified as Grade S or A, subject to manual verification; well-known media and institutional domains are rated by manual inquiry and comparison with similar sources; completely unrecognizable sources are classified as Grade E.
D.6 Full Source Library Description
This appendix only lists excerpts from the authoritative source library. The full source library contains approximately 5,000 domain and publication entries, covering seven major global regions and more than 50 countries and territories, including government agencies, international organizations, authoritative media, academic platforms, industry standards bodies, third-party evaluation institutions, and high-quality community platforms. The full source library is regularly updated and maintained by the Source Review Working Group, and the latest version can be obtained through official Pangpei Index channels.
Appendix E: Data Sources and Disclaimer
E.1 Data Sources
The data and information covered in this white paper are derived from the following channels:
E.1.1 AI Model Output Data
The core data for WACI evaluation is obtained by sending queries to mainstream global AI large models through standardized API interfaces. The specific collection method is: for each target city, a pre-designed set of standardized questions is used to send query requests to the target AI model via API interfaces, and the complete response text is automatically recorded. Each question is independently queried 3 times on each model (with intervals of no less than 24 hours), and the average of the 3 results is taken to reduce the impact of randomness in AI responses.
The selection of AI models follows the principle of "global coverage, regional balance, and type diversity," encompassing global general-purpose models, Chinese models, European models, Middle Eastern models, and East Asian models. The specific model list is continuously updated with the development of AI technology, and the current version of the model list can be obtained through official Pangpei Index channels.
E.1.2 Basis for Source Authority Determination
The source authority determination involved in WACI evaluation is based on the "WACI Authoritative Source Library." This source library is independently maintained by the Source Review Working Group of the Pangpei Index Expert Committee, covering categories such as government agencies, international organizations, authoritative media, academic platforms, industry standards bodies, and third-party evaluation institutions, with regular annual reviews and updates.
E.1.3 Basic City Fact Data
The basic city fact data involved in WACI evaluation (such as full city names, countries/regions, leading industries, representative enterprises, etc.) is sourced from official city government websites, public data from national statistical agencies, authoritative media reports, and academic research literature, serving as the benchmark for determining the accuracy of city information in AI responses.
E.1.4 Data Collection Cycle
WACI adopts a two-tier collection system of "annual major assessment + quarterly dynamic tracking." The annual formal collection is conducted twice per year, and quarterly dynamic tracking is conducted once per quarter. The data covered in this white paper was collected during the 2026 assessment cycle.
E.2 Data Description
E.2.1 Data Limitations
Generative AI responses have inherent randomness—the same question may yield different answers at different times or across different models. WACI mitigates this through multi-round sampling with averaging and cross-model validation, but it remains a source of measurement error. AI models' knowledge bases and algorithms are continuously updated, and model version changes may lead to systematic fluctuations in evaluation results. WACI monitors and controls this impact through quarterly dynamic tracking and cross-cycle stability analysis.
Additionally, AI models' training data exhibit uneven language distribution—English content dominates the training data, which may cause systematic bias in certain indicators for cities in non-English-speaking regions. WACI reduces this bias through cross-language calibration and language-weighted balancing, but it cannot be completely eliminated.
E.2.2 Data Refinement Notes
WACI's global city assessment work is continuously advancing. At the current stage, some cities have completed multiple rounds of systematic data collection, while data for other cities is based on a comprehensive analysis of initial collection and simulated assessments. As the data collection system continues to improve and the city sample gradually expands, more comprehensive and systematic evaluation results will be released in the future. The data and findings presented in this white paper reflect the assessment progress as of the completion of this white paper.
E.3 Disclaimer
E.3.1 Research Nature Statement
This white paper is a research report based on the WACI 2.0 assessment framework, independently compiled by the Pangpei Index Research Team. The analyses, judgments, and conclusions in this white paper are research-oriented and exploratory evaluations, and do not constitute a comprehensive judgment of the assessed cities' overall strength or governance level, nor do they represent the official opinions or positions of any government agency, international organization, or commercial entity.
E.3.2 Non-Decision-Making Basis Statement
The content of this white paper is for research reference and information exchange purposes only, and does not constitute investment advice, a basis for business decisions, or the sole reference for policy formulation. Any entity that makes decisions based on the information in this white paper and the consequences arising therefrom shall bear such consequences itself. Users are advised to make comprehensive judgments in conjunction with other information and consult relevant professional institutions when necessary.
E.3.3 Methodology Disclosure Statement
The indicator system, weight allocation, data collection methods, and calculation logic on which this white paper is based have been fully disclosed in Appendix A. The raw data of basic indicators and calculation codes can be requested through the official channels of the Pangpei Index, subject to independent verification and supervision by academia and the public.
E.3.4 Version and Update Statement
This white paper is compiled based on the WACI 2.0 model. As AI technology iterates and methodology optimizes, the assessment model will be continuously updated. Evaluation results from historical versions do not constitute final judgments; scores and rankings may differ across versions, and it is recommended to refer to the latest released version. Updates and revisions to this white paper will be promptly announced through the official channels of the Pangpei Index.
E.3.5 Appeal and Correction Mechanism
If assessed cities have objections to the data or analyses in this white paper, they may submit written appeals or supplementary materials through the official channels of the Pangpei Index. The Pangpei Index Research Team will conduct a review within a specified timeframe upon receipt of the appeal and handle it in accordance with unified rules. If data errors or factual deviations are confirmed, corrections will be made in subsequent versions.
E.3.6 Intellectual Property Statement
The copyright of this white paper belongs to the Pangpei Index Research Team. Without written authorization, no institution or individual may use all or part of this white paper for commercial purposes. Academic research and policy citations must indicate the source. Trademarks such as city names and corporate names used in this white paper belong to their respective rights holders.
E.3.7 Independence Statement
The Pangpei Index Research Team is independent of any assessed city and AI model provider. The compilation of this white paper does not accept targeted funding from any assessed entity or AI model provider that could influence the evaluation results. Compilation funding comes from the institution's own funds and public issuance revenue. The independence statement is publicly released annually.
E.4 Contact Information
If you have any questions, suggestions, or appeal requests regarding the content of this white paper, please contact us through the official channels of the Pangpei Index.
About the Pangpei Index
The Pangpei Index is an AI cognitive competitiveness evaluation system founded by Mr. Pangpei, with the PAI framework (Pangpei AI Index Framework) as its core methodology. Based on real output data from globally mainstream AI large models, it systematically measures the presence, trustworthiness, and discourse power of cities, brands, industries, entrepreneurs, and governance entities in the AI cognitive world. The Pangpei Index has released seven indices covering five major fields, committed to providing an open, transparent, and verifiable measurement benchmark for global competitiveness in the AI era.
Theoretical Founder: Pangpei
Research Institution: AI Index Research Institute / China Vision New Media GEO Research Institute / WICOWEB Global AI Cognition Research Center
Release Date: July 31, 2026