Every country has a different idea of what counts as progress. Taiwan, like Korea, believes in open innovation and the civic sector. The Beijing model, unfortunately, cannot make full use of the civic sector, journalism, or other distributed collective intelligence.

It depends on what you are using AI for. If it’s about authoritarian control, making citizens transparent to the state, social credit, and micromanaging the population, of course, they are unmatched in the world. They are running one way, and we, the free world, are running the other way. They ...

This is like when I was young, people thought of computers as mainframes—huge CapEx investments. But then Taiwan came up with the idea of IBM PC compatibles: personal computing. It’s a tiny fraction of what the mainframe does, but you can steer your personal computer, responding to you and your ...

I don’t think the US controls AI development. The US had a head start, of course, in applying the transformer architecture. But now people realize that less can be more. If you are translating, summarizing, or doing deep research on a mathematical puzzle, you do not need a huge model ...

This is sovereign, assistive AI in practice: resilient by design, privacy-preserving by default, and aligned with democratic oversight because everybody can steer it locally. It prioritizes a race to safety and trust over a race to larger, more monopolistic models. In my ongoing conversations with South Korean counterparts, this blend—cybersecurity ...

Imagine Samsung networks where local agents schedule power use across a neighborhood, optimize transit in real time, or deliver personal assistance and care for the elderly through relational health—without constantly exporting private data to the cloud.

The synergy is profound. Samsung provides the substrate — energy-efficient semiconductors, HBM for accelerators, and billions of edge “SmartThings” endpoints in phones, wearables, and appliances. Sakana supplies the coordination and evolutionary logic — agentic collectives that run on-device, preserving privacy, reducing latency, and shrinking the cloud attack surface.

This “small-but-smart” direction dovetails with Sakana’s agent swarms: many compact experts coordinating locally, enabled by Samsung research and hardware, rather than relying on some distant model in the cloud.

This vision pairs naturally with Samsung. Beyond its leadership in memory and devices—I’ve used the Galaxy Note and Fold for many years, since the original Note, because I use the stylus a lot and Samsung phones have some of the best—the Samsung Research Center has revealed in a recent paper ...

Globally, I was just in Japan meeting with a company called Sakana AI. They have a very particular model—not just artificial intelligence, but artificial life. They use an evolutionary, multi-agent approach. They don’t want one giant model; they want a collective intelligence of small, specialized agents that collaborate and evolve ...

I think so. You have, for example, the tiny recursive model from Samsung, which is open source. Honeybee from Kakao Brain is open source. There are many researchers contributing. Indeed, the open-source flagship product Linux was also sponsored by many large corporations, like Red Hat (now IBM). I don’t think ...

The benefit of horizontal alignment is like LEGO blocks: South Korea makes some pieces, we make some pieces, but because everything is open, we can build upon each other’s work. This commitment to fast, fair, and fun civic innovation is our most valuable export.

This turns AI into a tool for democratic resilience rather than a vector for top-down control, as the Beijing model is currently doing. This enhances the competitiveness of all participating countries by shifting the focus from possessing the largest AI to harnessing the collective intelligence of our people.

This approach counters authoritarian influence because it fosters a diffusion of diverse viewpoints. Because these models are open source, they can be small enough to run locally without continuous internet access. Then we gain true autonomy: the means to bypass top-down censorship from foreign places. The idea is to bring ...

The alternative approach, which we champion in Taiwan, is Plurality, or horizontal alignment. This means not a top-down sovereign deciding for everyone, but rather a civic approach. In Taiwan, we have 16 indigenous nations; they are all sovereign in some sense. It means empowering all those groups to fine-tune their ...

The isolated stack approach is sometimes called Singularity, or vertical alignment—the pursuit of ever-larger, centralized models. It requires massive capital investment using the transformer architecture, which exceeds what most nations can do. Usually, only the G2 can do this. I don’t think South Korea is trying to overtake the G2 ...

The point is that in an open ecosystem, it is not about creating isolated national tech stacks. It is about ensuring digital autonomy through collaboration on democratic values and, importantly, open-source principles.

My hope is that South Korea will export not just cutting-edge hardware, but also the values of openness and transparency in AI development, powered by the next generation of recursive, small, tiny model learning. By focusing on making civic AI possible, with community-tunable AI, our countries can collectively build a ...

The ability to steer AI is far more important than just for one country to press ahead with the accelerator; all countries in a democratic alliance need to be able to steer it together. As Samsung’s own research has shown, for specific tasks, smaller, specialized models are often sufficient. They ...

The key to security in the global value chain is not independence, but <em> interdependence </em> . When all democratic partners are indispensable to each other, we collectively raise the cost of disruption.

I also think cyber resilience and democratic governance are equally important. South Korea’s potential is immense, but to secure a long-term, trusted role—especially within alliances like Chip 4—we need to align our technological leadership with open governance, where no voice is left behind.

I think we are very complementary. It’s vital that companies like Samsung and SK Hynix are supplying the essential hardware that powers the global AI accelerators, just like TSMC is doing on the compute modules.

While TSMC excels in compute and chip manufacturing, South Korea is very strong in memory semiconductors, like high-bandwidth memory (HBM) and DRAM. Both are critical layers in the global AI value chain, the foundation upon which the entire AI architecture is built.

We also shouldn’t think of resilience as insurance that you pay in the hope it will get paid back sometime in the future. Cybersecurity investments must pay dividends along the way. For example, switching from passwords to passkeys is a security measure, but it also means you don’t have to ...

By collecting these credentials using zero-knowledge proofs, it doesn’t re-identify you. Because of that, the attack surface is much smaller. Even if attackers gain control of the system, they do not have sufficient information to retrieve this personal data.

This is why in Taiwan, we’re switching away from centralizing all personal information in one place. We collect credentials, not identifiers. For example, with our decentralized ID wallet, if you want to prove you’re over 18 years old, you can just prove you’re an adult without revealing your age. If ...

Everybody who holds massive amounts of personal data is vulnerable. The more centralized you are, the more attractive you are to the attacker.

A globally competitive AI ecosystem must prioritize an AI race to safety instead of just a race to capability. Concretely, this means focusing on assistive AI—AI that augments human agency—rather than just a centralized super-AI that holds our personal data. We involve citizens in AI regulation through open processes like ...

Centralized defense doesn’t work well anymore because it creates single points of failure. In an offense-dominant situation, that single point of failure becomes very attractive. Instead, we need to embrace decentralized resilience, where every attack strengthens our collective immunity.

My experience in Taiwan has shown that freedom and security are not a trade-off; they are mutually reinforcing. The key is adopting open-source principles and radical transparency. We need that in AI governance to cultivate a trustworthy environment where diverse perspectives collaborating—or Plurality—can flourish.

Without robust cyber resilience, advancements in AI can be easily co-opted to undermine societal trust. We see this materialize in the spread of deepfakes and misinformation, which erodes our shared sense of reality—our epistemic security.

My background as a civic hacker taught me that we are currently in an offense-dominant landscape. AI enhances the capabilities of attackers far more than it increases the capabilities of defenders. Using AI to find vulnerabilities or synthesize scams is easy; designing a system with no vulnerabilities is hard.

As Taiwan’s cyber ambassador, first digital minister, and architect of the Taiwan Model, I see the strategic importance of what we call <em> cyber resilience </em> as fundamental to preserving democratic values in the AI era.

And of course, we have, as I mentioned, meronymity and privacy-preserving architecture, so that when data is collected and processed on the edge or on-chip, you do not actually dox or re-identify people when you publish it into the commons to do sense-making. You just publish the differentially private aggregation ...

Part of the reason why is that it is not the individual checked facts from journalism or from the state that inoculates young minds against polarization. Rather, it is the act of going through civic journalism, reporting, and sousveillance <em> as a class </em> with their peers that inoculates the ...

In fact, that’s part of our basic education. If you’re a primary schooler in Taiwan, you learn about curiosity, collaboration, and civic care by participating in journalism, using sensors to measure air quality and water quality and hold presidential candidates to account as they’re having a debate in real time.

Yeah, that’s a great question. I think in Taiwan, there’s a lot of sousveillance going on. That is to say, people take live footage of the government and public sector doing things wrong in real time, too. It’s not just, as in some other countries, the state apparatus making the ...

I would also say that most people in Taiwan would not trust the government to run all this infrastructure. What we usually do is work with people like the g0v movement, so they also become a civic infrastructure. If you don’t trust the government’s wallet system for verifiable credentials, you ...

By giving public contracts for adjacent parts of the stack to different vendors, it forces the system to speak interoperable protocols. Then you can have civic AI, public AI, and so on, working with some foreign suppliers, but they can never lock you in. This is something I think the ...

The Ministry of Digital Affairs is uniquely in the Parliament’s transportation committee, instead of in science, technology, economy, or interior (for cybersecurity) as in other countries. This is because, in the transportation committee, people understand the idea that if you build an information superhighway, you must have some off-ramps. The ...

The really the first thing is to build such language models in the commons and make them widely available. Also, shape your policy environment such that if any part of the technical stack goes into public procurement, the same vendor cannot win the contract for the adjacent part of the ...

Okay, that’s a great question. In Taiwan, we focused mostly on, as I mentioned, the linguistic work across the public sector: summarization, topicalization, translation across our national languages, and so on. The thing is, if you know what you’re doing, you do not actually need huge models for any of ...

Sure. So just to clarify, you mean a country that wants to build the local AI stack without depending on any foreign power, and you’re asking what the initial things to work on are, essentially?

What Taiwan learned ten years ago, when we were literally at the frontline of political polarization and anti-social media, suddenly everyone is waking up to. We cannot really live in this high-PPM moment any longer. Conversely, there’s a lot of renewed interest and investment in democracy as a technology.

I’m feeling a sense of optimism because after last year—which I think had the highest number of people going to the polls ever—all the ruling parties lost seats. I don’t think there’s a ruling party that <em> gained </em> any seats. And they lost the seats not to the traditional ...

I think now people really do see that if we upgrade democracy for a higher bandwidth and lower latency, if we can make democracy work in the here and now instead of as this ritual we practice every four years, then we have a good chance of depolarizing society and ...

Well, I think democracy as a technology is really having a moment now, because people are collectively tired of polarization. No matter which side of the political aisle I talk to, when I visit the U.S.—the most advanced state when it comes to political polarization—both sides are collectively tired of ...

Sure. I think if there’s only one thing people can learn, it is that democracy can be fast, fair, and fun. You don’t need to make a trade-off. There are a lot of ways now, thanks to digital technology, that you do not have to just make democracy exciting for ...

What I’m continuing to do is essentially what I did since I was four years old: publish my half-baked thoughts online into the public domain so that human and AI systems can let the light in through the cracks. Instead of claiming I’m a perfect ancestor solving some worldwide problem, ...

Well, I think at this point, I’m just a pattern of thinking, right? It’s not Audrey Tang, this organic embodiment, but rather “Audrey Tang,” this way of prompting in the AI model and getting this exoskeletal way of collaborating across differences.