Thank you, Minister Tang. Do you hear me well and see me well?

Yes, I will make a voice recording of this conversation, so we can make a transcript. But we will not publish until you do. We will embargo, but this is just to preserve the context of our conversation for, you know, radical transparency.

Yeah, yeah. That would be also my first question, because normally when I have an interview, I ask my interview partner if I can record this to be safe of all the translations. And in your case, I don’t have to do this.

Yes, in a sense, we are very similar, right, providing context to the public.

Yes, yes. And it’s a kind of, yeah, something, it’s part of your political understanding that you record everything, that you transcribe it, upload it.

And my first question would be, what is your thought behind this radical transparency?

Yes, this is so that the people understand the why and how of policymaking, not just what of the policies made. This made it possible for people to understand the kind of trade-offs we’re making, as well as the context.

Because once you publish, the editorial control of the media, of course, will focus on the topics that the media publisher feels like focusing on. But more or less, this is a compression, and sometimes it makes the fuller nuance kind of lost, right? So, I think in the best of cases, this helps investigative reporters, because people who look at the journalism report and see that, oh, there’s something that is worth diving in, can go to our website and see the full context, and upon which they can also understand the why and how the thought process behind policymaking.

Is there something else of the radical transparency that you, for example, in governance, can use to show the people what you do and how you work?

But that is certainly a very important part of an informed citizenry, though my point is that because it’s rare that a citizen will just randomly read a very long transcript of a full interview or our monthly ministerial meetings and things like that, but it always helps the people working to write a story, a report, could be a journalist, could be a researcher from the industry or from the academy, and when they want to know, oh, so this is the policy, but why was it made? How was it communicated? And so, what they can refer back to.

So, in a sense, we’re not publishing scoops, which is why I’m comfortable saying we publish only after you do, because this is more for the historical purposes, for the people down the line who are affected by these policies to go back and see the kind of conversations we had, but this is not so that we livestream every meeting. This is not for the real-time purposes.

When Western countries look at Taiwan, we see you’re very forward in digitalization, and it’s not only because of the transparency, but also because of citizen participation.

Can you give me an example for Taiwan? What did you change in the years when you are now minister for citizen participation?

Yeah, certainly. I think part of Taiwan’s main countermeasure against not just the pandemic, which is also co-created, but also, the infodemic - foreign interference and information manipulation. In many other jurisdictions, we see the state assuming a more censoring role or a surveillance role, and it becomes, if you’re a journalist or media practitioner, sometimes it creates this tension where the journalist says something, but the state, because it’s not agreeable with the state’s ideologies or things like that, then restricts the journalistic freedom in the name of countering information manipulation.

But in Taiwan, the journalists are held in very high regard. My parents are both journalists. And so, because of that, we say it is the journalists and the citizen journalists that are the antibodies, so to speak, against foreign information manipulation. So, instead of teaching the students digital literacy, which is about reading, consuming, we teach them instead digital and media competence, which is how to have a journalistic outlook and help on fact-checking, helping, for example, when the three presidential candidates were having a platform or debate, the high school students can correct their problems in their statements and even appear their name on national live stream TV and so on if they find something wrong and so on.

And there’s an ongoing ecosystem of people contributing such context. It’s like a Twitter community note, but in Taiwan it’s called Cofacts - collaborative fact-checking, where everybody can join. So, by people learning about journalism, they form a journalistic mindset, which is the antidote, the inoculation against information manipulation.

There’s also, at school, the students learn how to fact-check, you said. And how important is this for such a young democracy like Taiwan that they have to fact-check and have this digital media competence?

Definitely. And this year, because of generative AI, the Cofacts project even used their previous clarification and fact-checking text to train their own language model, their bot, working with advanced transformer models. These are the kind of first-line clarifications.

So, if you go to the Cofacts website and see the trending disinformation and its countermeasures, oftentimes you will see a bot providing the first-time response. And this solves a critical issue against foreign interference because the foreign information manipulators are professionals. They’re paid to do this, like, nine to six days a week. But the students’ fact-checkers, of course, work in their part-time. Although many, they’re not full-time working at this. But now they can help train a language model that works 24-7 to provide the first-line of defense.

OK. So, this is also a step into artificial intelligence. And Time listed you as one of the most influential persons in artificial intelligence. It’s crucial because there are not a lot of politicians in the list of 100, and no minister at all. I think there’s a lot of fear, especially in Europe, about AI, that it will be dangerous for democracies. And many politicians focus on shielding the democracy from AI. What is your approach in there?

As I mentioned, we believe in inclusive co-creation. Co-creation means that everybody can train their own AI for their community’s purposes. Because exactly as you said, the current issue with the monoculture, the largest AI models, was that it provides the same answer, the same thought, no matter which community asks it the same question. But many issues are very nuanced. It responds to… For example, in Taiwan, we have more than 20 national languages. We have 16 indigenous nations with 42 language variations. And the same question may mean very differently to these different nations, different people.

Now, what we are essentially saying is that it needs to be closest to where the people are. The people need to have ways to not just run their own AI, like the way that I, on my MacBook, as we speak, run the local AI, that not just transcribe what we have said, but also. it can actually draft my responses based on my previous emails and things like that. But it’s entirely running on my laptop and not on the cloud. And it’s not using any of those cloud services, so it works on the edge in a non-connected way. So, this is individual.

But for the community, we also do what we call alignment assemblies, which we gather in a city. We’ve done this in Taipei and Tainan. We ask people for a month online to see what they feel about generative AI. And we invite the people who raise very good points to come face-to-face for a whole day and consolidate what they expect of a norm, right, what’s normal for AI to respond. And once we do that, you can check out the website. It’s called Talk to the City. You can actually have a conversation with those people’s avatar, so to speak, and then have a deliberation with the record of such online and also offline conversations. And then with that, we can train new AI models based on the norms that are gathered in this way.

And so, I think the main point I’m making is that first, it needs to belong to the people at the edge, closest to where people are. And second, it needs to be democratically aligned, meaning that people should be able to train it themselves and in a scalable way. So, I recently published a blog talking about this, which I pasted here.

And in Europe, there are a lot of people who are very sceptic about AI. So, how about the Taiwanese people? Are they very into digitalization? And is it also because of… It’s because of this very young democracy, which has to shield itself against the mainland China?

I think part of our positive outlook is that we have successfully countered the threats posed by the previous engagements of AI. For example, the social media recommendation algorithm is a kind of AI, right? It’s not generative, but it chooses what kind of posts will get the most clicks, will sell the most advertisements, will keep people glued to their phone. And it caused not just addiction, which cures over time for most people, but also polarization, which doesn’t cure itself. People get more and more divided.

But in Taiwan, thanks to the kind of collaborative fact-checking, literacy turned into competence, education, resilience, we successfully overcame the polarizing effect of social media AI on our society. So, this time, when faced with generative AI, for example, our cabinet passed a regulation, a guideline on public sector use of generative AI. And unlike, I would say, most jurisdictions, we encourage the public servants to use generative AI and to train their own generative AI, but with two caveats.

First, it must never replace people. It must be assistive, not automated. And second, if it pertains to personal data, then the entire running and training and everything need to happen off the internet. It needs to be done, as I said, on my MacBook, closer to the edge, not connected to the cloud.

So, this kind of personal computing or assistive intelligence is why we have this positive outlook, and maybe because we make most of the chips, I guess. [laughter] So, this is not a mythical thing for us. This is just a product that our supply chain makes.

OK, OK. And we have seen in 2016, there were some Russian disinformation campaigns that influenced the US elections and the US election campaign. There will be a presidential election in January in Taiwan, too. And do you expect such disinformation campaigns from Beijing also? And how do you shield yourself, especially for this big event?

Yeah, it’s not specifically leading up to any big day or something. According to the Secretary General of our National Security Council, Wellington Koo, every day we face, like, five million attack attempts. So, it’s not like there’s no cyber-attacks, and closer to the election we get some, we get millions every day. And so, because of that, I think we’re also very experienced in defending against this kind of cyber-attacks.

We already adopted what we call a zero-trust posture, meaning that we’re shifting away from passwords. Passwords are very easily phished or scammed or things like that. And after they gain the username and password, they can escalate its privilege and then post hate messages or some other stuff on the website and so on. So, phishing passwords is often the first stage in having a cyber-attack that leads to manipulation, right?

But, like, this year, all our ministry, our underlying administrations are completely password-free. So nowadays, when I want to use official systems, I use this phone that checks my fingerprint, but my fingerprint stays on the phone. And another software checks the phone so that it’s not changed or replaced, and yet another software checks the behavior of the phone.

And so, these three doors, they don’t belong to the same company. So, because of that, even if the cyber-attack compromises one of the three doors, it still gets stopped by the other two doors, and they’ve wasted an attack, because then we will share the threat indicator with the Democratic Alliance.

So, can you say, is China flooding, or is the mainland China flooding Taiwan with…

We say foreign manipulation and foreign cyber-attacks because we’re a bunch of islands, right? So, we can tell whether an internet traffic is domestic from within our jurisdiction or external, that’s to say, travel to Taiwan via subsea cables, for example.

But where did it originate outside of Taiwan? Did it go through what’s called botnet or zombie network, whether it goes through Tor or VPN or things like that? That varies. So, to be scientifically accurate, we just make a distinction between domestic and foreign interventions.

OK. And they always say Taiwan is the country with the most cyber-attacks. Is it true? Do you have some relations to that, to other countries?

Yeah, I think so. It’s certainly one of, if not the, right, the hottest spots. If you look for the varieties of democracy or V-Dem research, the V-Dem research has consistently placed Taiwan on the top, under the kind of threat. It’s, I think, a Sweden-based institute. So, if you search for V-Dem Taiwan, you will find the quotes.

Mm-hmm. OK. And when Nancy Pelosi visited in August last year, there were also very heavily cyber-attacks on Taiwan. On the screens of 7-Eleven, the supermarket chain, you could read, War Monger Pelosi Get Out of Taiwan. How hard was it for you to shield Taiwan against the foreign cyber-attacks?

Yeah, I mean, that was just a few weeks before the Ministry of Digital Affairs started last August. And we, of course, saw that it was highly coordinated, right? It first disrupts the websites. It’s called denial-of-service attacks. And then it posts the hateful messages and so on, as you mentioned. And when the journalists want to check the official clarification of what actually happened, they find the website very slow. They cannot connect to it, which also adds fuel to the information manipulation.

So, the cyber-attack arm and the information manipulation arm, they work in tandem, very coordinated, last August. The goal, of course, was to make our people not believe in democratic institutions, to make our people panic, to make the stock market crash, maybe. But on that day, the stock market actually grew, so they did not achieve what they were originally trying to achieve. And more importantly, it’s very wasteful of their resources, because in that single day, we suffered 23 times more DDoS attack volume, and it’s all very expensive, compared to the previous most, the peak day, right? So, it’s like unprecedented by 23 times, the kind of resource they threw at denying our website. And so, we immediately kick into action.

And this year, although we saw comparable DDoS attacks and other assaults, it did not achieve the kind of results, as you said, last August anymore, because, as I mentioned, we switched to zero-trust defense systems, as well as asking people to help keeping us afloat, right, by making sure that we use the content distribution networks and also IPFS, which is a way for anyone, even people in regimes that are not democratic, if they want, they can run IPFS to help backing us up using their own hard disk and network.

Okay. And we’ve seen this also in the Ukraine. Since years before the attack on the whole country, Russia launched a lot of cyber-attacks, especially on the infrastructure. There were some blackouts in 2014. And what is your way to shield Taiwan against such an attack from Mainland?

Yeah, we learned a lot from the Ukrainian experience. As you know, Ukraine also has more than one telecom provider, and they enabled what we call disaster roaming, right? So that even when your SIM card is registered to a telecom and that telecom’s infrastructure is gone, you can still connect to other telecoms that usually don’t provide service, but in terms of disasters, they do provide service to people needing network.

And we also learned from Ukraine in working with non-terrestrial networks, specifically non-geostationary satellites. And so, before end of next year, we will have more than 700, some fixed, some mobile, satellite receivers in Taiwan to provide emergency communication, especially to journalists and disaster response teams if our subsea cables are cut. So, I think both in the satellite sense and also disaster roaming sense, we did learn a lot from Ukraine.

There’s also the Kuma Academy in Taiwan. I visited last year, and November, and yeah, the people learn how to be prepared for such an attack. How are they going to be prepared for such an attack in the digital way? Is there something they can learn for this moment if it would happen?

Yeah, I think one of the main strengths in Taiwan is that we have a very, very strong civil society. So, when the civil society people, like the ones you mentioned, finds, for example, Ukraine has this new way of doing things, they immediately get into people-to-people contact and exchanges with the DIA team, the team in Ukraine doing that super app, or people in Estonia who have been also preparing this for decades now.

And because many of these are what we call public code, meaning that the software and the policy that enable the software are out there, like, transparently. Estonia very famously shared the X-road system with Iceland and Finland together. So, this is a democratic alliance maintaining the same software public infrastructure to defend against this kind of attack.

So, when our civil society gets into contact with these people, they don’t have to, like, send an official MOU or something, because it’s all out there in the open, and it’s literally just some URLs, some hyperlinks. And once they develop this, we call it civic technology, then our ministry also works with the civic technologists to ensure that they receive the kind of support they need in, for example, translating such civic technologies, in educating the public sector about the existence of such civic technologies, and so on. So, I would say that Taiwan is one of the most vibrant places for civic technologies.

Mm-hmm. You also helped Ukraine, and I think in the first days of the attack, is there still a connection to Ukraine? Do you help them build up the digital resilience?

So, the DIA team, before they repurposed DIA for this purpose, they actually already were part of the Open Government Partnership, and we had met, like, the ProZorro system and so on. That was just all before the illegal and brutal Russian invasion. So, there were already some connections, as you mentioned, between the civic tech community in Taiwan and the people in Ukraine, and that connection, of course, deepened once that we noticed that we face very similar threats, especially on the network resilience space.

Mm-hmm. There was, just last week, I think, yeah, it was Elon Musk who said that he prevented a Ukrainian attack on the Russian fleet. He said he did not agree the request because SpaceX would be explicitly complicit in a major act of war. And how do you see this, that, yeah, the provider interferes in a war, and do you think that this could be a problem for Taiwan, too?

One way we learned from the Ukrainian experience was that there’s a strength in plurality. So, right now, as we speak, the more than 700 satellite receiver project, the emergency resilience project, the 700 spots are specified to have more than one satellite provider as backup.

So, not just the existing geosynchronous satellite, but also middle-Earth orbit like SES, or low-Earth orbit like OneWeb, but we also say, you know, the more, the merrier, right? If one spot has three, four, five different satellites, and if the data and the system that we need running are backed up in three or more public cloud providers, then an adversary will be extremely resourceful, required to destroy all those satellites and all those public cloud providers. But if we just use one provider with one satellite provider, yes, it is possible, actually, to be disrupted.

OK. Would you say Taiwan is already in a cyber war with the mainland?

I would say, as I mentioned, quoting Wellington Koo, Secretary General of our NSC, there’s more than five million cyber-attack attempts every single day from foreign sources.

OK. So, you don’t want to be very, very clear, but…

No, no, it’s just scientifically, we only know that these come from outside of Taiwan, from the subsea cables. Its IP address may show that it belongs to this jurisdiction or that jurisdiction, but we also know that the attackers use botnets across different jurisdictions. So, we wouldn’t simply, because the IP address is in this jurisdiction, to say, oh, the attack comes from this jurisdiction, because that would be oversimplification.

OK. And Taiwan, there are currently just 12 countries in the world that recognize Taiwan as a country itself. How can digitalization improve the relations with other countries and perhaps improve the recognition of Taiwan as a country for itself?

In the Declaration for the Future of the Internet, the DFI, there are more than 60 democratic partners, and Taiwan is one of them. So, instead of saying that Taiwan is a jurisdiction, slash area, slash country, slash self-governing island, or whatever, we simply refer to ourselves and each other as democratic partners, because this is a democratic alliance. The alliance members are naturally partners or democracies.

And so, this is a new form of hybrid multistakeholder and multilateral relationships. Our membership in the DFI, our membership in the Democracy Summit Declaration, in many similar recent arrangements, simply joining as a full partner or the full democracy.

And something a bit more personal, there’s hardly no article about you which doesn’t tell about your extraordinary life story. You taught yourself programming when you were eight, you dropped out of school when you were 14 because you wanted to focus on the research in the internet, and you also founded, I think, some IT companies with friends.

Can you talk a bit about what shaped your life path, what shaped your life decisions to be such connected with the world wide web?

Yes, certainly. So, I remember when I was 14, 15, I talked to the head of my school, Principal Du Huiping, and I said, “Madam, I don’t want to go to school anymore.” And she’s like, oh, did anyone hurt you, were you bullied? And things like that. And I’m like, nothing like that. It was just that you told me, Principal, to study hard, to get a good degree, to finally study abroad, joining a top AI lab maybe, working as postdoc to a professor and so on. But now there’s this new thing called the World Wide Web. And on the archive website, I just write an email to such professors, and they don’t know I was just 14, and they start sending me emails, and we start working together. And it’s like I found this fast track that I can do research full time without having to complete the diploma or credentials required for research.

And because that was very early days of the World Wide Web and of archive, everybody was very open, very eager to make new friends. Still nowadays, ARXIV, hosts the leading edge, not just artificial intelligence, but many other fields of research. So, the head of my school said, yeah, tomorrow you don’t have to go to school anymore. I see your point. I read the email printouts. And without her blessing, because it’s compulsory education I would not be able to start a company so soon and lead to the free software and open-source community that you mentioned later. So that very much was a watershed day for me.

So, you started your company when you were 14, is that right then?

So, I quit school, quit attending school when I was 14, but I started my company, co-founded really, when I was 15.

15, okay. That’s a very young age. Did the people who worked together were older and were they surprised?

Yeah, they’re much older. They’re 25.

Okay.

Yeah. 15 is important for legal purposes because in Taiwan, child labor legally begins at 15.

Oh, okay. And what fascinated you about the World Wide Web at such a young age? I know when I was young, the internet was mostly for some play and Google. And what was the thought behind it that you think was so fascinating about the Internet?

How quickly people trusted complete strangers online was the thing that fascinates me. It’s called swift trust, right? Because face-to-face, my parents would say, don’t go to strangers’ homes. If the stranger gives you a treat, don’t take it, and so on. I mean, there are social norms in the face-to-face world.

However, because across the screen, you cannot abduct me anyway, right? So, the harm is limited across the screen. And because of that, people come to trust the people they just met across the screen and talk about the most intimate and personal things because they know that it’s harder across the screen for them to hurt each other. Of course, that was before the social media and cyberbullying and broadband. I was talking about the early internet, right?

And so that really fascinated me and how the researchers, as I mentioned, are willing to share, even before they publish to the journalists, their ideas with completely random people they met across the internet. So, I made it my research focus to focus on what makes the swift trust and also distrust. Like, we can very quickly hate people, unfriend people across the internet too, much faster than if we have to meet face-to-face, right?

And so why people quickly trust each other and distrust each other, that becomes the main fascination.

And it also is connected to democracy, I guess, so this trust. And you said because of social media, do you use social media yourself for personal purpose or are you more afraid of using it because of the harm it could take?

Yeah, so I use social media in two conditions. One is that I never touch the screen because when I touch the screen, very quickly, it becomes the screen controlling my finger, not my finger controlling the screen because they were designed by a persuasion technologist that are designed to make addiction easy across touch screens.

So, ever since the days of Palm Pilot, which is long ago, I’ve always used the stylus on the screen, right? So, this screen is actually also a tablet and I always interact with it through a stylus or a mouse or a keyboard, but never a touch screen. And I find that this is really the only way that enable me to just say, okay, I’ll just put down my phone. I don’t bring my phone outdoors, right? I am not addicted to it because I don’t touch the touch screen, first thing.

And the second thing is that on, say, Facebook and so on, I don’t have a newsfeed. I use a software called Newsfeed Eradicator that removes the newsfeed. So, I use Facebook like a blog. I post something, a video or some thoughts. I go search, look for hashtags and visit people and have interactions, send messages, but no AI controls what I see. So, I remove the newsfeed. And once the newsfeed is removed and I don’t touch the touch screen, I find it to be a very pleasant place.

Okay. And, yeah, a question because I’m also German and we are now here in the middle of Europe. You spent some years in Saarbrücken. And can you talk a bit, when was it? And did it like… yeah, what is your impression of it now all these years after?

Yes. I had a very good education and experience in Dudweiler near Saarbrücken because it’s close to the France (34:30-34:33) border. So, we had a bilingual education actually. And it’s very nice because even though I don’t speak any bit of German or French when joining, still at a very handicapped place to start, the people were very nice to me and they trusted a child as if the child is an adult. This is very different from the Taiwanese culture back then.

In Taiwan, the kids stay kids in their family until at least 18, but sometimes older. But in Germany, I was 12 when I was in Germany and my classmates were 11, I believe. No, I was 11 and my classmates were 10. But the adults are treating those 10-year-olds just like other adults. And this is a Pygmalion effect, right? Once the children are expected to keep the time, to be punctual, prinklish, to be orderly, in ordinal, and so on, they learn to be very mature. So even though the 10-year-olds are technically one year my junior, my classmates, when compared to the kids I meet in Taiwan, they seem older by at least four or five years.

So, this had an impression on me in that, oh, actually the children can make important decisions concerning themselves and we should trust children to join also in public participation before they turn 18 because if you shield them from the democratic and civic process before they’re 18, they don’t magically become informed citizens when they turn 18.

And this is possible in Taiwan that they can… [be] connected more to the civic participation before 18?

Exactly, exactly. So that’s why we made it a focus to switch from literacy, comprehend something, to competence, make something. Education, I was part of the basic education curriculum committee that in 2019 have this new curriculum that focused on the sort of common good, autonomy and interaction, competence-based education.

So, I would say, yes, we changed our education just in time for generative AI because anything that requires memorization and comprehension and whatever probably can be delegated to personal AIs and it is the creativity, the autonomy, the common good, the competence that makes sense now.

Yeah, I think it’s very funny because here in Europe we just started the discussion how do we cope with AI, with Chat GPT and children who copy-paste their homework and so, Taiwan is there more forward.

Yeah, I think we’re very forward now. We’ve all embraced generative AI and as I mentioned, we focus on edge AI, local AI that can be tuned, trained by the students and also ones that can be democratically aligned to fit the social norm.

Yes. My last question would be, we already talked about the elections in January. Is there a chance that you will be nominated again for digital minister or do you have some other plans after that?

Well, I’m a double appointee, as you know, so people directly elect the president but the president appoints the premier or the head of the cabinet and then that person assembles their cabinet.

So, it is not for me to say and also it is not yet for any of the presidential candidates to say because the cabinet head, the premier, is not part of the ticket. We vote for president and vice president but they don’t say who will be the head of the cabinet. So once that person is determined, which is likely sometime between next January and next May, then that person will say, oh, I want my cabinet to look like this.

Okay, and would be a chance that you would do it again for the next term?

We now have not just three but potentially four presidential candidates. So, it is far from me to say that, oh, I think this candidate, that candidate, they will nominate this person as head of cabinet. There are just too many variables for me to comment in a matter-of-fact way for this question.

And after your, I don’t know if it’s after your political career, do you have some other plans? I think you have a lot of prospects.

Yeah, as I mentioned, I am working with Taiwan, of course, but I’m not working exclusively for Taiwan. I am, after all, an e-resident of Lithuania and also a digital resident of Palau, among other things. And so, I think we, in the democratic alliance, work in a way that transcends simple jurisdictional boundaries.

I think the previous president of Estonia, Ilves, said it best. He said that on the internet, there’s no geopolitical neighbors. It’s not determined by geology or geographics. He said that it is the shared values that binds us together. So, people who share democratic values are neighbors because everything travels on the speed of light anyway, right? And people who are autocracies that are attacking the democratic polities, well, they also share their playbooks together between them. So, it’s not a geographic neighborhood. This is the value, a shared value neighborhood.

And I think I already work very closely with many Europeans as advisors or as board members and so on that affects responsible AI and digital transformation of many jurisdictions. And I will continue to do that, no matter what my association with the Taiwan jurisdiction is.

Okay, okay. Minister Tang, that would be all of my questions. Thank you very, very much for talking to me. It was very interesting. And thank you so much for having the time to talk to me. And I hope we see you soon and talk soon again. And I will…

So, when roughly will you publish? Do you know?

Not exactly.

Sometime this year?

Yeah, yeah, of course. I think in the next two to three weeks, I guess so. But I’ll keep you informed, of course. And I will show you and share with you the article. I will write that.

That was a very nice conversation. Thank you for the very thought-provoking questions.

Thank you very much. So have a nice day.

You too. Live long and prosper. Bye.

Bye.