Conversations that cross borders
where we listen, speak, and understand.
Public Data for Trustworthy AI
1) Problem Recognition While organizations like the OECD, UNESCO, and NIST have established comprehensive AI ethics and governance principles (inclusivity, human rights, transparency, accountability, etc.), execution models specifically addressing “how AI accurately reflects each country’s public information” are still in their early stages. In particular, global AI models frequently distort or oversimplify the policies, institutions, history, and cultural information of non-English speaking countries due to language and regional biases. 2) International AI Information Management Principles (Draft) Principle of Prioritizing Public Data For information regarding public policy, institutions, and administration, each country should provide public data and official documents as “ground truth for AI training,” and AI developers must prioritize the use and reference of this data. To achieve this, governments must open public data in machine-readable and AI-friendly formats, while mandating metadata that specifies the source, responsible agency, and update cycle. Principle of Source & Evidence Transparency When AI systems answer questions regarding high-risk information (public policy, laws, statistics), they must provide information allowing users to verify “what type of data (public data, laws, international statistics, etc.) was used as the basis,” to the extent possible. Even if providing a direct link is difficult, the system should be recommended to provide metadata describing the material type, issuing agency, and timestamp. Principle of Inclusive & Non-Discriminatory Expression When providing information on nations, regions, or groups, AI should aim for narratives that align with the OECD AI Principles of “inclusive growth, human rights, and fairness.” To minimize expressions that reinforce prejudice, hate, or colonial perspectives, a “Sensitive Expression & History Distortion Checklist” should be developed in cooperation with international organizations and civil society. Principle of Joint Verification & Correction Responsibility A principle stating that the responsibility for monitoring and correcting AI information errors is shared not only by AI developers but also by: National Governments (Public Data Providers) International Organizations (Standard & Guideline Setters) Civil Society & Academia (Monitors & Verifiers) 3) International Cooperation Models Establishment of the ‘Trustworthy AI Public Information Working Group’ (Based on OECD/UNESCO) Create a new “Working Group on AI Reliability for Public Policy & Administration” based on existing international standards such as the OECD AI Principles and the Recommendation on the Ethics of Artificial Intelligence. Roles: Compare and analyze public data/open data policies of each country to suggest minimum common standards suitable for AI training and verification. Collect and share best practices (e.g., Korea’s AI Ethics Standards, the EU AI Act, Governance Frameworks of Canada/Singapore). Establishment of the “Global AI Error Observatory” An online platform operated jointly by international organizations (e.g., OECD, UNESCO) and member states. Functions: Build a database of AI information error cases by country and region. Classify and publish statistics on error types (factual errors, bias, discriminatory expression, misinterpretation of laws/statistics). Record correction status and follow-up policy/institutional improvements for each case. Utilization: Member governments, companies, and civil society can use observatory data to supplement domestic public data, improve agreements with AI developers, and design education/media literacy programs. “Public Data for AI Trust” Joint Campaign Goal: Globally spread the message that “AI trust begins with high-quality public data and citizen verification.” Key Activities: Each country announces plans to open and standardize public data tailored to their language and local context. Host international hackathons/challenges where citizens, researchers, and the press report AI information errors and verify them against public data. Allow participating companies to use a “Public Data-Based Trustworthy AI” logo or certification, inducing competition for trust in the market. 4) Expected Effects Forms a new axis of cooperation internationally connecting public data and AI governance, contributing to the reduction of information bias that favors specific countries or languages. Korea and Gyeonggi Province can position themselves as leaders in the field of “Public Data-Based AI Reliability” by sharing their experiences with AI Ethics Standards, the AI Basic Act, and public data infrastructure.AI-driven heritage distortion
Translation: The Importance of Critical Thinking in the Age of AI AI technology has advanced at a staggering pace in recent years. In just two or three years, the level of technology has improved remarkably. As a result, unlike in the past when generative AI was prone to many errors, many of those issues have now been corrected, and people have become increasingly reliant on the information AI provides. However, AI-generated information still frequently proves to be imperfect. A clear example of this is the result of asking an AI to generate an image of an East Asian "Samo" (a traditional official’s hat). The generated image differs significantly from the "Samo" we know. Furthermore, we can see that the text describing the "Samo" often contains incorrect or nonsensical characters. This serves as a cautionary tale: no matter how rapidly AI technology develops or how much its error rate decreases, we must not blindly trust the information it gives us. Despite this, we see many people around us accepting AI-generated information without any critical filter. This ranges from providing information to others without cross-referencing to submitting university assignments by accepting AI output as-is without verification. Such an attitude can lead to serious consequences, such as deceiving people by presenting falsehoods as truths and eroding our own capacity for independent thought and critical analysis. Therefore, I propose a solution to prevent such cases of distortion. While petitioning to correct the information provided by AI is important, I believe it is even more vital to change our fundamental attitude toward AI. It is crucial to fix the habit of indiscriminately accepting information just because it is convenient, even though it may not be accurate or true. To achieve this, schools and educational institutions should teach people how to use AI in a healthy way without being controlled by it. We must learn to use AI as a supplementary tool rather than relying on it entirely, and to use it while cross-verifying information with other sources such as books. I believe this will help AI users utilize the technology effectively without compromising their thinking and critical skills.Mandatory Education on African
1. Problem Statement Despite Africa’s rich history and diverse cultures, it is often portrayed in generative AI platforms and educational materials primarily as a region of poverty and hardship. This one-sided representation reinforces stereotypes and limits students’ opportunities to understand Africa as an equal partner in global cultural exchange. Such bias stems from insufficient cultural education and the uncritical use of AI-generated information. 2. Policy Objective This policy aims to correct biased perceptions of Africa by integrating balanced cultural education into school curricula and promoting critical engagement with generative AI platforms. The goal is to foster global awareness and mutual respect through education. 3. Policy Recommendations First, introduce mandatory education on African history and culture in primary and secondary schools. African civilizations, art, music, and contemporary cultural movements should be included in social studies and world history courses, shifting the focus from poverty-centered narratives to culture- and exchange-oriented perspectives. Second, implement project-based learning to encourage active student participation. Students will research an African country of their choice and create digital content—such as posters, videos, or card news—highlighting its history, culture, and modern society. This approach promotes deeper understanding beyond memorization. Third, incorporate critical AI literacy education. Students should be taught to analyze and evaluate AI-generated content related to Africa, identifying stereotypes, generalizations, or inaccuracies. This will strengthen their ability to use digital technologies responsibly and critically. 4. Expected Outcomes Through this policy, students will come to view Africa not merely as a recipient of aid, but as an active cultural and historical contributor to the global community. In the long term, this initiative will help build a more balanced and respectful global perspective.Real Africa: Beyond AI Bias
Many people still perceive Africa mainly through stereotypes such as poverty, disease, and conflict. These images are repeatedly reinforced by media narratives and even by generative AI outputs. In this forum agenda, I would like to discuss how AI and online information may misrepresent Africa as a single “problem continent,” and how global society can promote fair and balanced representation. We can share real examples of misinformation, explore the causes of bias, and suggest practical solutions such as education reform, responsible media content, and AI bias reduction guidelines.“One Nation, One Voice in AI”
To safeguard trust in AI-generated information together with the international community, we propose a global campaign in which every nation declares the principle: “Each country defines the official standards for information about itself.” Under this initiative, each country will: Create an official AI reference page covering its history, culture, and policies Link this page to international platforms Encourage AI companies to prioritize these pages for training and reference In partnership with UNESCO, the campaign will include: An exhibition of national “AI Standard Declarations” A global youth project where students create their country’s official pages The publication of international reports comparing AI errors by country This model: Reduces information structures dominated by powerful nations Ensures that every country has an equal right to tell its own story Transforms AI from a “controlling technology” into a “tool that connects the voices of nations” In the age of AI, a new form of diplomacy will emerge based on the shared principle: “One Nation, One Voice in AI.”Global AI Data Campaign -GADC
Globally, AI-generated information often contains inaccuracies or biases based on preconceptions. For instance, information about Korea and Gyeonggi-do may be misrepresented or perceived through biased perspectives by foreign users, and similar issues exist in other regions such as Africa. To address this, the “User Participation-Based Global Trusted Data Campaign” encourages users from each country to provide accurate information about their own region and culture directly through an app. Participants answer simple questions and, in return, receive tangible incentives such as image generation or shortcut features. This approach minimizes regional bias and misinformation, enabling AI to learn from more reliable data while simultaneously promoting user engagement and awareness. The expected outcomes include enhanced accuracy and trustworthiness of AI information, improved cultural and regional understanding, and the establishment of a global participatory data ecosystem.prevent distort heritage in ai
If you search for Korean cultural heritage using ChatGPT or Gemini, you will find something completely different. It creates or creates images with the influence of Japan and China, which are close to our country. No matter how long and detailed the prompt is used to explain, AI cannot recognize it because it is still under development. But I don't think it should happen that Japanese and Chinese cultural heritage or letters go in. So, when you search for cultural heritage using the ChatGPT or Gemini, you develop an algorithm that doesn't mix it with the Chinese/Japanese culture. I think we should protect our cultural heritage in the digital world. The attached picture is that I created an 'SASINDO', which is a cultural heritage of our country, through AI. Even so, I was heartbroken to see that Chinese characters were written, not in Korean language. So, I suggested a policy because I thought it would be good if I could search differently from the nearest country first.Responsible AI for Heritage
Artificial Intelligence is becoming a major tool for generating cultural content and promoting heritage across digital platforms. However, the absence of shared global standards has led to inaccuracies, mislabeling, and cultural distortions in AI-generated materials. These cases do not only harm cultural integrity, but also risk creating diplomatic misunderstandings and misinformation in the international community. To address this challenge, we propose the establishment of an international framework for responsible AI use in cultural heritage. The framework may include: Cultural Data Verification Protocols Collaboration with cultural institutions, museums, and academic bodies to provide verified cultural datasets for AI training and inference. Global Heritage Label Guidelines Standardized naming, contextual description, and origin labeling for heritage-related AI outputs. Monitoring and Reporting Mechanisms Crowdsourcing channels for citizens and experts to report inaccuracies in AI-generated cultural materials. Capacity-Building & Shared Platforms Joint programs helping AI developers and cultural agencies understand heritage sensitivities and avoid cultural bias. This proposal aims not to restrict innovation, but to foster responsible AI adoption, safeguard cultural diversity, and support accurate global cultural exchange. We encourage cooperation among governments, cultural bodies, AI companies, and international organizations such as UNESCO and WIPO to co-develop and refine this framework.AI Heritage Network
In the era of globalization, AI technology can be a powerful tool for promoting cultural heritage, but it also carries the risk of spreading misinformation and distorted representations. To address this challenge and promote responsible interaction between AI and cultural heritage, we propose the Global AI Cultural Heritage Trust Network, involving collaboration among nations, institutions, and private companies. The goal of this policy is to ensure that AI-generated cultural heritage content accurately reflects historical facts and cultural significance, providing reliable information for users worldwide and enhancing the global promotion of cultural heritage. Key initiatives include the development of international AI cultural heritage verification standards in partnership with UNESCO, the International Council of Museums, and leading AI companies, as well as the operation of a joint platform to evaluate the accuracy of AI-generated content. Verified content will receive an international trust certification mark for use in online promotion, education, and tourism campaigns. Additionally, a global participatory platform will allow citizens, tourists, and researchers to report errors and contribute to content improvement. Finally, educational programs and social media campaigns will promote the responsible use of AI-generated cultural heritage information worldwide. By implementing this policy, the spread of distorted AI-based cultural heritage information can be minimized, and a globally trusted system for cultural heritage promotion can be established, increasing both the international recognition and credibility of cultural heritage in Korea and other countries. 세계화 시대에 AI 기술은 문화유산 홍보의 강력한 도구가 될 수 있지만, 동시에 잘못된 정보와 왜곡된 이미지가 빠르게 확산될 위험도 존재한다. 이에 따라 국제사회와 협력하여 AI와 문화유산의 올바른 상호작용을 촉진하기 위해, 국가와 기관, 민간 기업이 함께 참여하는 글로벌 AI 문화유산 신뢰 네트워크를 제안한다. 이 정책의 목표는 AI를 활용한 문화유산 콘텐츠가 사실과 문화적 의미를 정확히 반영하도록 국제 표준을 마련하고, 전 세계 이용자가 신뢰할 수 있는 정보를 공유함으로써 문화유산의 글로벌 홍보를 강화하는 데 있다. 주요 내용으로는 첫째, UNESCO, 세계박물관연합, 주요 AI 기업과 협력하여 국제 AI 문화유산 검증 기준을 개발하고, AI 생성 콘텐츠의 정확성을 평가하는 공동 플랫폼을 운영한다. 둘째, 검증된 콘텐츠에는 국제 신뢰 인증 마크를 부여하여 온라인 홍보, 교육, 관광 홍보 등 글로벌 채널에서 활용되도록 한다. 셋째, 시민과 관광객, 연구자가 직접 오류를 신고하고 개선에 참여할 수 있는 글로벌 참여형 플랫폼을 구축한다. 마지막으로, AI 활용 문화유산 교육 프로그램과 SNS 캠페인을 통해 전 세계 사용자에게 올바른 정보 활용 문화를 확산시킨다. 이 정책을 통해 AI 시대에 문화유산 정보의 왜곡을 최소화하고, 국제적으로 신뢰받는 문화유산 홍보 체계를 구축함으로써, 한국 및 각국 문화유산의 글로벌 인지도와 신뢰도가 동시에 상승할 것으로 기대된다.