-
You can see, when I’m speaking, there’s sound, but when you speak, there’s none.
-
(interviewer speaks)
-
Oh, yeah definitely. When people talk in the morning, I guess there’s no prior engagement, like this conversation is my first engagement. My thinking is quite fresh. It’s like a blank canvas. At evening, I’m burdened by other things that I’ve heard throughout the day.
-
(interviewer speaks)
-
I do think so.
-
(interviewer speaks)
-
Definitely.
-
(interviewer speaks)
-
I prefer… whatever. I don’t have a real preference. [laughs]
-
(interviewer speaks)
-
During the height of the pandemic, it’s 10, 12 calls a day. Nowadays, it’s maybe just 8 or something. Early morning for me connects to the East Coast of the American people, and the evening connects to the African and European.
-
(interviewer speaks)
-
That’s right.
-
(interviewer speaks)
-
Usually, it’s with think tanks. Some of them bipartisan, some of them more closely affiliated, one party or the next, but all the conversation so far has been quite bipartisan in nature. Everyone want to of course counter the pandemic. Nowadays, it seems everyone want to counter the infodemic.
-
(interviewer speaks)
-
Glen Weyl, who you also know. [laughs]
-
(interviewer speaks)
-
That’s right. It’s more about infodemic, about how to build data coalitions, data trust. That’s his day job at Microsoft.
-
(interviewer speaks)
-
Not particularly. Glen has arranged quite a few meetings with people who are more into testing that, of course, I gladly participated, but otherwise, it’s more like journalists. There was one pretty good NPR episode, that’s also in their podcasting, so quite a few news coverage about the work.
-
I think I’m more working with the media community, and individual politicians or scientists want to learn more. We usually communicate through email, so less synchronous communication.
-
(interviewer speaks)
-
It was a pretty good write up. That’s actually a one-time interview, but I think the author really did a lot of extensive interview, investigative too, with other people in the civic tech network.
-
(interviewer speaks)
-
Sorry, which foundation?
-
(interviewer speaks)
-
Well, through email certainly, but usually if they ask about specific measures, I just refer them to the specific teams within our center for epidemic command.
-
That has the linkage to the broader scientific community and so on, which as you mentioned in the article, is still globally cherished community despite political absent flows and so on. I usually refer them to the team of epidemiologists. I’m not an epidemiologist. [laughs] I just connect to the epidemiologists.
-
(interviewer speaks)
-
That’s right.
-
(interviewer speaks)
-
Probably that will be very helpful. I can connect them to their counterparts in Taiwan. I’m not involved in the day-to-day preparatory work for, as we call it here, SARS 3.0, but I do know that people are preparing for SARS 3.0. The world is seeing maybe on 2.5 beta or something at the moment.
-
(interviewer speaks)
-
Not really. I was very heavily involved back last February to maybe June. That was around the securing the supply and distribution fair equitable access to the PPEs and things like that, but now there’s a…I mean we don’t really have any community spread. There’s really not much work to be done.
-
Whenever there’s a few local cases, of course the contact tracing, the medical officers are very busy, but from a information technology viewpoint and infodemic viewpoint, all the mechanism is in place.
-
I’ll probably be back to work if there’s a community spread, but it doesn’t seem like likely, or when the vaccination is ready to roll out, then maybe like the health care access and things like that, then that would need some work. Otherwise no, I am not involved in the CECC’s daily work now.
-
(interviewer speaks)
-
There’s a batch, the first quarter, just enough to protect the health workers. There’s the second batch around June, some of which are also domestically produced.
-
That plus the COVAX, plus I think some direct purchases, should cover us because we do have a physical vaccine anyway. That’s the prevailing use of masks. I think the R-value stays below one, nevertheless, but I think we’ll get a second batch rolled out around June.
-
(interviewer speaks)
-
Yes. We have two in phase 2 and with EUAs, the rollout can be combined with phase 3 because it’s showing pretty good scores and so on in phase 2.
-
(interviewer speaks)
-
Of distribution.
-
(interviewer speaks)
-
Yes, I’ve thought about it and we’ve got certain plans. Specifically for the distribution, but also for a domestic productions as well.
-
(interviewer speaks)
-
Again, I am happy to connect them to the CECC because CECC is the team that comes up with the requirements for cybersecurity parameters and so on. I am mostly just in charge to make sure that the checks and balances are in place to make sure that these contractors and so on fulfill their cybersecurity parameter requirements.
-
(interviewer speaks)
-
I am still at Digitalminister.TW, so yes.
-
(interviewer speaks)
-
There’s nine ministers without portfolio in the Taiwanese cabinet and each of us have our ad-hoc portfolio because “without portfolio” means that there’s no fixed portfolio. I’m still the Digital Minister in charge of social innovation, youth engagement, and open government.
-
(interviewer speaks)
-
Yes, without portfolio means without fixed portfolio, for example, social entrepreneurship is also my portfolio even though it is somewhat connected to digital, I guess.
-
(interviewer speaks)
-
Yes, youth engagement and things like that.
-
(interviewer speaks)
-
The other what?
-
(interviewer speaks)
-
Yeah, there’s nine, or as I refer to as horizontal ministers. For example, we have minister with a portfolio, John Deng, who is the chief trade negotiator.
-
(interviewer speaks)
-
Definitely, yes. So all of us span multiple, at least five or more ministries in our work.
-
(interviewer speaks)
-
Before joining the cabinet, I worked with the Siri team for six years.
-
(interviewer speaks)
-
It all depends on the norm. In the norm that I referred to as assistive intelligence, like any assistive technology, it’s by default aligned like my eye glass, being assistive technology aligned to my interest of seeing better and not displaying pop-up advertisements.
-
The design of the eye glass industry is so that when it breaks, I can go to the nearby shop to repair or even fix it myself without paying astronomical licenses and fees.
-
It all depends on the norm. There are places where the norm is more authoritarian, which I refer to as the authoritarian intelligence, where people enslave other people through AI. In Taiwan, it’s definitely assistive in the sense that it’s democratic and people can learn, having access to very high speed GPUs as part of the K-12 Curriculum anywhere with broadband as human rights.
-
People learn to repair assistive intelligence and devise new ones from a very tender age. That makes it assistive by nature. We do not endorse or support the use in public service, the things that are not aligned nor accountable.
-
(interviewer speaks)
-
Yes, that’s right. Opaque eyeglasses are not very useful. Let’s call it a blind or something. [laughs]
-
(interviewer speaks)
-
Yeah, definitely. For example, the Taiwanese equivalent of Reddit is the PTT. It’s a good example, because it’s co-governed. It’s operated by the social sector, subsidized entirely by the National Taiwan University and has no shareholders, no advertisers.
-
The “regulatory body” is just people who are participating only for a long time, with too much time on their hands, which is equivalent to the Wikipedia governance.
-
All the AI algorithm that’s deployed in the PTT is vetted through the open-source development system. The technical part is held in check by the co-governing regulatory part that is comprised entirely of the social sector, which in other countries is called civil society organization or volunteer sector, depending.
-
That gives it a really good bargaining leverage point, when we talk with Facebook, for example, demanding for algorithmic transparency when it comes to political advertisements and micro-targeting, we can say ‘If you don’t do this…
-
We are doing it from a national auditing office, the PTT part is doing it. You may face social sanctions because there’s already a strong people-public partnership. What we want is people-public private partnership.
-
In 2019, Facebook deployed in Taiwan, as the first jurisdiction in the world, the real-time advertisement library that banned foreign interference via sponsorships and so on. This is just one example. There’s many like this.
-
(interviewer speaks)
-
Glen’s colleague, Divya, I think, did a research paper on it, also from RadicalxChange. I can send it to you afterwards.
-
(interviewer speaks)
-
Divya did extensive research on existing materials.
-
There is a really good one about using AI to regulate AI, that’s the UberX example where we using the Pol.is system, which is a AI moderated facilitation method so that uber drivers and taxi drivers and all passengers can reach a rough consensus on how to regulate the UberX algorithm. That’s well worth a read.
-
(interviewer speaks)
-
The system is called “Pol.is”, and the process back in 2015 is called vTaiwan. V for Virtual.
-
(interviewer speaks)
-
All the democratic regimes have now a tendency to become more authoritarian either in response to the pandemic or in response to the infodemic or in response to whatever other threats. The more top-down, or take-down, lockdown, shutdown it is, then the less chance it is for the social sector to develop the kind of norm that is required for this kind of co-governance.
-
I’m cautiously optimistic in democratic countries but it requires a certain sense of the…like the term social sector itself. It requires this kind of term that unites together the people who wants to build public infrastructure in the digital realm.
-
(interviewer speaks)
-
I’m at Digitalminister.TW slash, as you know, RadicalxChange board member, slash CONSUL Foundation board member, that’s in Amsterdam, but started from the Spain 15-M movement. The CONSUL of Democracy Foundation bands together the kinds of networks and people across all countries that think about these things together.
-
I’m also slash Digital Future Society. It’s a think tank associated with the Mobile World Capital, Barcelona. They just launched this Digital Bill of Rights.
-
That’s after extensive consultation. I think the Spanish government decided to basically put what we just told them in writing and advocate for that sort of governance within EU.
-
(interviewer speaks)
-
There’s a chapter about demanding the right to not be automatically judged essentially, and also to judge automatic systems and see how much we’re comfortable with them judging us individually.
-
(interviewer speaks)
-
Sure, yes.
-
(interviewer speaks)
-
Yeah, it has a nuclear feel to it. Isn’t it?
-
(interviewer speaks)
-
That’s right. Yes. That’s right.
-
(interviewer speaks)
-
That’s right. Put a millisecond cap to it.
-
(interviewer speaks)
-
To check my understanding. You worry mostly about the high bandwidth and low latency decision making assistance that are enabled, if not by this current generation of thermo semiconductor chips, then certainly by the next. People will delegate warfare to these algorithms. There is currently no international…Not even norm, let alone non-proliferation that regulates these?
-
(interviewer speaks)
-
That’s fine.
-
(interviewer speaks)
-
I am.
-
(interviewer speaks)
-
A little bit. There’s “self-driving cars are on average safer” arguments. I have heard of this before, but it sounds very authoritarian, to be frank. Only authoritarian designer of policies and systems which get by with this very top-down blend economy arguments.
-
In reality, in the polity that I’m more familiar with, it could be Taiwan. It could be other Taiwan-like countries. The Sidewalk Labs/Canada case is a really good case. What I’ve found is that on average better utility doesn’t really fly in a democratic polity and then people would demand to be much more informed before yielding so much mandate to these systems.
-
Now, I understand in matters of defense that community may look at it differently, but I also see many people in the defense or intelligence community worrying about precisely the same thing you worry about and arguing precisely in the way that you say that we may be unconsciously captured, for lack of better word, by these authoritarian imaginations.
-
I think as long as there’s sufficient amount of people holding this reasonable fear, uncertainty and doubts, I don’t see that decision makers, I mean presidents and so on, investing too much into this unconscious buyers when called outs, buyers is no longer one.
-
(interviewer speaks)
-
I understand and totally agree, yes.
-
(interviewer speaks)
-
It’s a great analogy, because elevators were called cars. [laughs]
-
(interviewer speaks)
-
I totally agree. The point I was making is purely that once the elevators – thank you for the analogy – becomes self-driving cars on rails, because elevators operate through certain tracks and self-driving cars, at least in Taiwan, I think people become accepted it only in scenarios where it’s software defined rails in a certain sense.
-
That basically makes it very clear that if you get in front of the car, that’s certainly your problem, not the car’s problem.
-
The civil society organizations, of course also popularize the algorithms. Middle schoolers, college students can all build such “self-driving cars,” which are following the standard computer vision systems and so on. People learn to be familiar with and demand accountability whenever any accidents happen.
-
We’ve got maybe five different test beds for self-driving because integrating with the society, what I refer to as co-domestication going around. I totally agree. At some point, it will be co-domesticated.
-
What I’m trying to say is that people do not give up their resistance purely because of, on average, it kills less people calculus. They are less resistant because they’re more familiar and also understand more how it works, posing accountability but also in alignment.
-
(interviewer speaks)
-
Certainly. And wolves/dog is great for my analogy of assistive intelligence, because dog is a assistive intelligence while it’s not a artificial intelligence. [laughs] A lot of metaphors carry over very easily.
-
Of course, I don’t know much about the psychology of dogs, but I know people who know. I know people who can teach me a little bit if I want to know more. The same goes for elevators.
-
It’s a ladder of expertise in the society that makes sure that whenever we need access to justice, when the dog or the elevator behaves erotically, then this justice in a epistemic sense, but also in a lower sense is achieved.
-
(interviewer speaks)
-
I know, yes. [laughs]
-
(interviewer speaks)
-
Everybody seems very comfortable with it. [laughs]
-
(interviewer speaks)
-
I think that’s the different pedagogical view. One view is so-called literacy. If you take this view, then if you’re surrounded by literate people, then whatever the social norm is, well is, and our job will be then to make sure that people are literate in media literacy, or elevator literacy, fire literacy, dog domestication literacy, [laughs] and so on and to accept the ongoing norm.
-
In Taiwan, our education system underwent a pretty large change a couple of years ago and shifted from this literacy view to the competence view mostly in response to AI, but also to other emerging digital phenomena where the students are no longer passive consumers of norms because they fact-check their teachers all the time. We need to adapt to that.
-
We teach instead media competence where they are the producers and need to be aware of the social externalities – positive or not – of their actions as producers in media and also digital ones and so on, which is a much more as you put it a inquisitive view on education in general.
-
I don’t think there’s a lot difference between you and me because a decade ago in any classroom in Taiwan, your view will be dominant. It just so happen that I work on the new curriculum. We’re actively steering the society to a competence view.
-
(interviewer speaks)
-
A little bit. I usually quip saying, if you have an access to a smartphone, not even the Internet, with five apps, everyone has a IQ that’s immeasurably high because it’s measuring the things that apps can do for you as assistive intelligence is. It’s not important at all long as you know how to use those apps properly, or even right one.
-
I think it’s a great equalizer much as a self-defense, martial arts or weapon. It’s a great equalizer across different body conditions. A app is a great equalizer for IQ, for anyone who have to access can pass the IQ tests with the top score and flying colors.
-
(interviewer speaks)
-
Yeah, well, to my understanding, it’s much more easy access to knowledge in the form that’s more amenable to rational thought. I’m not just talking about Wikipedia or broadband as a human right, because we know in many corners of world, we don’t have problems of the human rights.
-
Just the fact that people understood that science is not just about preparing some papers and so on, but about bite-sized knowledge that could nevertheless be replicated in day-to-day.
-
Popular science, for example, discovered many ways for us to understand the essence of things without understanding the theoretical knowledge of things, and so on. That is my understanding of why the rational thought capability has been risen in places that are exposed to this kind of pedagogy.
-
(interviewer speaks)
-
That’s right. Yeah, the epistemic lens. Yes.
-
(interviewer speaks)
-
Exactly. Because if you’ve put it in the frame of resistance, like giving consent, then of course familiarity, peer pressure, social norms, that’s dominant. My initial frame was that is answering to the utilitarian calculus argument, through an epistemic perspective. The main difference is how you frame this.
-
I can see people complying but feeling slightly anxious, accepting it in a social-pressured sense, but not in a intuitive gut-feeling sense. That’s probably describing many people going into hyperloop for the first time?
-
(interviewer speaks)
-
That’s right.
-
(interviewer speaks)
-
We might be both correct. Regulatory technology, supervisory technology, these are also technologies. Democracy itself is a type of technology. The point is to increase the bandwidth of the regulatory technology.
-
When the accelerating returns from the AIs, that you just mentioned, stops to be assistive and become addictive intelligences or evolves from being symbiotic to being purely parasitic.
-
Regulatory technologies and technologist that works on the technology. Thus democracy needs to take action. That’s how we reined in, for example, addictive anti-social social media. We’ve had some experience in the civilian world.
-
(interviewer speaks)
-
It’s very much isomorphic, even in both the cyber-security, which knows no borders by default, and also on climate change, carbon dioxide is like this. It knows no borders and it spreads out of control. We developed some regulatory tech, but at much slower pace.
-
(interviewer speaks)
-
Definitely, optimism takes two forms, right? One is that, well, we get to Mars. Mars doesn’t have a planetary climate change issue. We’re going to change the Mars climate before getting there anyway. That’s a very easy escape route. I’m not offering that sort of optimism – I guess there’s more of that in your country? [laughs]
-
What I’m offering instead is a working example of a smaller polity that nevertheless, I have balanced speed of social sector-led regulatory technology, innovation, social innovation, as we put it here, and take a social sector first norm.
-
When it comes to emerging technologies, including AI and self-driving vehicles. The fact that Taiwan’s polity exists as such and has been quite successful actually in spreading these ideas around our corner of the world shows a viable alternative.
-
(interviewer speaks)
-
Definitely, even just on the cyber-security side — which is a intermediate point — actors may accidentally shut down the Internet connectivity, and then what?
-
(interviewer speaks)
-
Bureaucracies operate in the timescale of minutes. Before high frequency trading, market were in the seconds and dogs, of course, works in milliseconds as human beings are. Of course, if we can or regulate the pace of AI in decision making to be only milliseconds or seconds level, then I completely agree it would be the same lesson. That’s a separate question.
-
(interviewer speaks)
-
That’s right. If it’s low enough latency, then I think we can only ask for justice after the fact instead of during the decision making.
-
(interviewer speaks)
-
That’s right, and when everybody does so, then I think we avoid the entire inadvertent first strike, whatever issues, and that’s clearly the sane thing to do.
-
(interviewer speaks)
-
In your article that I read, it was really moved by the examples of polio vaccination and things like that, that the Russia and the US collaborated in smallpox, I believe.
-
I just want to raise a point that in early January last year, we from Taiwan sent two medical officers to Wuhan and collaborated with them on assessing the human-to-human transmissibility of the virus. That’s even before the lockdown and all that.
-
(interviewer speaks)
-
I think it is a closer, temporarily speaking example, I just want to bring it to your attention.
-
(interviewer speaks)
-
They get pretty good access. They shared in the debriefing back because we heard from social media, from Dr. Li Wenliang that there is, and I quote, “Seven new SARS cases.” The two experts were there basically to check Dr. Li Wenliang’s message, which did reach pretty much everyone in Taiwan, but pretty much none of the people in Wuhan.
-
They have to, of course, investigate by themselves, but they do get the…not really evidence, but anecdotes that they believe confirm what Dr. Li Wenliang was saying. That informed Taiwan’s opening of the Central Epidemic Command Center even before we had the first local case and certainly before the lockdown.
-
(interviewer speaks)
-
In Taiwanese social media, yes.
-
(interviewer speaks)
-
For the government, we started health inspections for all flights, passengers coming in from Wuhan on the 1st of January, which is the standard protocol. We didn’t know for sure whether it’s human-to-human transmitted. We wrote to the WHO, which didn’t write back. Because of that, we need to see it for ourselves.
-
I think it’s a pretty positive sign that still Wuhan people did not refuse our experts, and the fact that our expert did get a sufficient amount of anecdotes or circumstantial evidences that they report back. That led to the CECC being opened just in time for border control.
-
(interviewer speaks)
-
Well, afterwards, there’s no way to get to Wuhan anymore. They literally locked down the city. That’s one shot. Of course, by the time the Wuhan city unlocks, everybody knows about SARS 2.0 anyway.
-
(interviewer speaks)
-
SARS wasn’t asymptomatic. There’s a lot of circumstantial thoughts about not just asymptomatic, but also about airborne. The more the virus mutates, the more likely that some variant will be transmitted more easily in aerosols and things like that. All that took much more time to suss out. I don’t think they list this as more than possibilities.
-
(interviewer speaks)
-
No, I really don’t.
-
(interviewer speaks)
-
Well, I think for a PRC audience, I think what would make them happy would be recognizing more their authoritarian model of, they call it 举国体制 or “nationwide” model of extreme authoritarianism for compliance, things like that. That’s precisely the kind of thing that is very difficult to transmit outside of their regulatory model and to establish as a new norm.
-
You’ve been wise in essentially not mentioning that.
-
(interviewer speaks)
-
Not for the moment, but I really enjoy our conversations around elevators and dogs. [laughs]
-
(interviewer speaks)
-
Yeah, sure. Just send me an email.
-
(interviewer speaks)
-
Yeah, thank you, and let’s bring this work to international communities, which I understand is facing a much larger challenge.
-
(interviewer speaks)
-
OK, all right. That’s it for now, then.
-
Live long and prosper. Bye.