Pioneering the Future of Think Tanks: Launching 'EAI Global,' an AI Agent-Based Policy Knowledge Platform

YouTube Link: https://www.youtube.com/watch?v=g5GFkb04uwc

* English subtitles/CC are available for this video. *

Video Script

The Launch of EAI Global and Its AI-Powered Policy Report Production System

Today, I would like to introduce EAI Global, a new initiative from the East Asia Institute (EAI). Founded in 2002, EAI is a private think tank that operates several platforms, including its main homepage, Global North Korea, the Asia Democracy Research Network, and our publication, the Journal of East Asian Studies. The newly launched EAI Global is a new site that will publish in-depth reports on pressing daily issues, distinguishing it from the more long-term, in-depth reports traditionally published on the main EAI homepage.

This is a multilingual site published in eight languages—Korean and seven others. The global order is currently undergoing fundamental changes, and EAI has been producing a variety of research reports on different issues and regions to keep pace. However, we have come to believe that our existing division of labor is no longer sufficient to fully grasp the transformations in the global order, as all regions and issues—military, economic, social, and cultural—are now interconnected. We have therefore spent much time considering how to comprehensively understand these rapidly evolving issues. We believe this new site presents an opportunity for us. Although we are a small private think tank with limited research staff, we can now actively leverage emerging artificial intelligence agents to investigate and organize information about events happening worldwide. The entire process involves our institute's researchers and their associated networks,

with researchers collaborating at each stage. We are embarking on this new endeavor because we have entered an era where AI agents can handle tasks that previously required significant effort and time. Over the past year, we have worked diligently to prepare for this, aiming to fundamentally transform the knowledge production system within the policy ecosystem. We plan to publish various types of reports daily, targeting a very diverse readership. Above all, we believe this site will deliver more accurate information to various officials in the South Korean government and provide valuable perspectives and facts to the media, corporations, civil society, and fellow policy researchers.

We believe that this attempt to actively use AI agents to produce policy reports is a pioneering effort globally. While we cannot know what results it will yield, we created this video in the hope that our viewers will read and utilize our research reports in various ways. We think this will not only serve South Korea's national interest but also become a completely new paradigm for producing policy knowledge in the age of AI. As will be explained in more detail later, we intend to analyze events happening around the world by rapidly collecting information from local websites and official government documents as they become available. Crucially, this analysis will fully reflect the research achievements and diverse perspectives that EAI has cultivated over the past two decades.

Therefore, these reports are quite different from simple AI-generated summaries, as they incorporate the theoretical analysis and wisdom EAI has developed, from the planning stage to data collection, analysis, and policy recommendations. We have worked hard to implement this technically. We hope you will follow the reports we provide with interest. Some content will be available on a subscription basis. We are promoting this initiative with the hope that you will utilize the reports produced by EAI Global in various forms according to your needs. Next, an advisor to EAI will speak in detail about the technical foundations, implementation process, and applications of EAI Global. He played a comprehensive role in the technical implementation of this endeavor, so we will be sure to hear from him.

Our chairman, Ha Young-sun, will also speak about what policy knowledge means in the age of artificial intelligence and how the role of think tanks must change. We will listen closely to his remarks. Next, we will hear from our advisor, Gong Ri, who will introduce the EAI Global site. Yes, it's a pleasure to be here. I have been brainstorming with the researchers about how to apply AI to the institute, considering the overall status of the EAI Global site.

The Research Function of AI Agents and the Role of Human Researchers

The most critical and time-consuming stage of report production is research. We focused on the fact that applying AI to this stage can significantly broaden the scope of investigation while reducing the time required, providing researchers with high-quality research materials in a short period. For EAI Global, we have the AI explore local sources from 192 countries worldwide. It scans over 600 data sources daily to identify the most important global and regional issues of the day. A human researcher then intervenes to direct the AI on the perspective from which to investigate the selected issues. The AI then produces broad, in-depth, and, most importantly, locally-sourced research findings on the given issue in a short amount of time—in as little as five minutes. We call this output a "preliminary research report." Based on this, a human researcher then adds depth

Measures to Ensure the Accuracy and Reliability of AI Reports

and refines and supplements it to create the final research report. In other words, the AI prepares extensive research findings in a short time, and the human researcher adds depth, makes additions, and refines the material to complete the final study. Another area we focused on was the "hallucination" phenomenon, which many people are skeptical or worried about. To address this, we technically "caged" the AI by creating a vector database that forces it to answer only from within the collected data sources.

It does not go outside these bounds; it cites, reinterprets, and rephrases content only from the sources it has collected. Furthermore, since it could misinterpret dates within that data, we added a system to automatically verify dates. We also implemented a process that accurately adds footnotes for all citations and rephrasings and then automatically verifies that those footnotes are correctly placed. Based on our checks and reviews so far, we consider it a major achievement that we have secured the integrity of dates and, especially, citations. The reports created this way are then published, and this too is powered by AI.

Simultaneous Multilingual Publication and Global System Development

The reports are published simultaneously in multiple languages for a global audience. We are publishing concurrently in eight languages, including Korean, English, Spanish, and Arabic. After much consideration and numerous tests regarding translation quality, we are now publishing simultaneously in these eight languages. The site is configured so that if your browser or smartphone is set to English, the site will appear in English, and if set to Chinese, it will appear in Chinese. We have established a comprehensive system where human researchers write reports based on the results of extensive research into local materials, and these reports are then published globally in eight languages. Internally, this involves vast processes, numerous verification mechanisms, and many procedures, but in broad terms, this is the production and publication system we have created. Thanks to your intensive efforts over the past year, it has become possible to produce reports that fully utilize AI agents.

A Pioneering Attempt at an AI-Powered Think Tank and the Rise of the "K-Think Tank"

I see this as a remarkably new and systematic development. On one level, EAI is celebrating its 25th anniversary, but I believe this represents the first expression of another, qualitatively different leap forward from our past. I don't think this work is just for EAI. Statistics show there are about 6,000 think tanks worldwide. Among them, EAI appears to be the first to operate as a think tank powered by AI agents in a truly comprehensive sense. For example, institutions like Brookings, CSIS, and the RAND Corporation are also making efforts, but compared to their work on AI agent-based report writing, which has been very rudimentary or limited, our attempt is considered a much more significant step forward. Third, looking a bit further

ahead, if this work becomes fully established and successful, I believe it could leave a new mark on the world as a kind of "K-think tank." As you know, the emergence and evolution of AI are bringing us new civilizational standards amid a historic transformation. Therefore, the success of this attempt could be a small but globally significant experiment. From the perspective of a researcher, a user, South Korea, or the world, I have personally wondered: what happens if one doesn't pay attention to this?

Expanding Understanding of the Global Order Through AI Collaboration

First, I believe a change like this is coming. As was explained earlier, as someone who has studied the global order my entire life, I have always wondered how to best approach it. As humans, we see the world through very limited news and media. In contrast, the work EAI is attempting involves intensively reviewing nearly a thousand sources, with a focus on over 600. Of course, this is practically impossible for a human. When the internet revolution first happened, the excitement was about getting closer to the reality of international politics and the global order. It was an incredible dream to move beyond just reading news from wire services and newspapers to engaging with the world's communication mechanisms directly. The experiment we are witnessing today is not about a human reading all this alone, but about reading and tracking the content through collaboration with AI. Therefore,

I believe it will create a situation where those who do not see or hear this, who do not read these analytical reports, will be living in a vastly different world. Second, what I find interesting are major recent events in world politics, such as the conflict between the United States and Iran or the processes of leadership change in Latin America. Behind the scenes, these were made possible with the help of AI agents that have undergone tremendous intellectual evolution. For example, tasks like setting targets, prioritizing them, and ultimately executing a plan would be utterly impossible for humans alone. In our case as well, when the AI reads over 600 important sources 24/7 at a speed no human can match, it's not just reading; it's reading in collaboration with us. This is why, while people who follow the global order have typically focused on what Putin is saying or how the UK's Brexit issue is being handled,

or what Chinese think tanks are saying, we are now, for the first time, able to look at these issues from an EAI or a Korean/Asia-Pacific perspective, examining the data sources more completely and figuring out how to apply that knowledge. To me, this means we are looking at a medium that provides an absolutely greater understanding, one that is not inferior to what other think tanks offer. And perhaps the framework that filters this information will also be seen as far more sophisticated at EAI.

Seeking Policy Alternatives for Korea and the Asia-Pacific

Finally, and this will be reflected in the content of the reports, our work is not primarily about just global affairs or issues. Rather, I believe EAI has begun to provide valuable insights that could not be obtained from American, Chinese, European, or Third World think tanks—answers to the question of what policy alternatives can be sought for Korea, the Korean Peninsula, and the Asia-Pacific world. This may sound like self-praise, but I believe that long after this launch in August, it will be seen as a small civilizational turning point. The story will be that the first pioneering experiment in the global think tank community—not just for a peninsular think tank—began on the Korean Peninsula. Lastly, I think this will achieve more than expected in helping many domestic think tanks, policy practitioners who need brainpower, and many others, including those present here, to think in a more sophisticated way.

The National Interest and Global Value of EAI Global Reports

With that, I will conclude today with a round of applause. Starting today, EAI Global will publish several reports daily. As mentioned earlier, these reports are co-produced by AI agents and our institute's researchers. While individual authors may be listed and recur, the reports in fact reflect the perspectives of multiple authors and the many scholars who have contributed their insights to EAI. Our research reports are beneficial for the Republic of Korea's national interest and its research ecosystem, but because they are delivered in various languages worldwide, I believe they can also hold shared global policy value. Therefore, we hope and expect that our reports will not be just for a single think tank or for Korea alone, but will serve as an excellent foundation for collectively achieving the global peace and prosperity we all desire. As these reports are automatically translated into various languages, there may be translation errors.

Therefore, we would appreciate it if our international readers could provide comments on any errors they find in the translation process. We will also strive to minimize translation errors and ensure our true meaning is conveyed. We would be very grateful for any other feedback you might have on areas for improvement. Thank you.