FT Weekend Fest: Tech to save democracy

0:00Audrey Tang: AI has opened infinite possibilities for customer experience and endless complexity because piloting AI tools is easy. Making them work together across a global enterprise is something else entirely. Nice is the CX AI platform that runs customer experience at scale, uniting agentic agents, human agents, and customer intelligence. So the world's leading brands can move forward with confidence. Nice, the proven CX AI leader. Learn more at nice.com.

0:35Hello, I'm John Thornhill, the FT's Innovation Editor. Tectonic will be back soon with new episodes. But in the meantime, here is another highlight from the FT Weekend Festival, held last month in London. It featured lots of interesting on-stage discussions around technology and AI, including mine on how technology can help revive democracy. In particular, I wanted to look at this idea of participatory democracy and how the internet, smartphones and AI could allow citizens to take a more direct role in governing their countries. So here's the discussion. Good afternoon. I'm John Thornhill, the tech columnist at the Financial Times, and it's a great pleasure to be hosting this discussion on how technology can help revive democracy.

1:25and you'll notice that that is a statement not a question because we are going to have a very positive vision about how technology can be a help which will make a bit of a change because I think the standard discourse today on technology is very dark. Social media has created filter bubbles and political polarization. AI threatens to drown us even further in AI slop but we're going to be talking today about how technology can fulfill its original promise of informing and empowering people. And for that discussion, we're joined by Audrey Tang, who is a Taiwanese software programmer and originally a student activist, who then served eight years in government, becoming Taiwan's first digital affairs minister in 2022.

2:10And Audrey is now Taiwan's roving cyber ambassador, promoting the concept of participatory democracy. So Matthew Rycroft is a former civil servant and diplomat who ran the Home Office and was Britain's ambassador to the United Nations. Since retiring from the civil service, Matthew has helped launch the National Strategy Project in Britain, which has drawn a lot of inspiration from Audrey's work in Taiwan. So Audrey, I'm going to start with you. Tell us, how did you first become involved in digital activism and how that led on to you becoming a government minister?

2:45Audrey Tang: Certainly. So in 2014, the Taiwanese president, Ma Ying-jeou at the time, in a country of 23 million people, was enjoying 9 % of approval rates, which means that whatever the president says, 20 million people against him. But those 20 million people themselves are extremely divided. Everything is just polarized into one extreme or the other. One side want to be free from China forever. Another side want to be the free China forever. And they seem pretty incompatible. At that time, the government was fast-tracking a trade agreement with Beijing that would have allowed their investment into our telecommunications, cybersecurity, publishing industries.

3:25Audrey Tang: But the people felt that this process had no social license to operate, meaning that the people affected had no say in it. And so we occupied the parliament peacefully for three weeks with half a million people on the street. But we always say we're not protesters, we're demonstrators. So we're actually doing tables of 10 deliberations on all aspects of their trade deal with people on the street and many more online. And amazingly, the idea that people can come to uncommon ground, that is to say the ideas that convince people of opposing side of ideology, really stuck with people. So after three weeks, we managed to convince each other and converge on a set of very workable ideas for the social license to operate for Huawei and France.

4:07Audrey Tang: And then the parliamentarians, the head of the parliament at the time, Wang Jinping, simply said, okay, the people's ideas are better. So let's just pass those ideas and go home. And so that kind of almost anarchic kind of system then got institutionalized, as it were, by the government. And they brought you into the tent. So what did you achieve when you were in the government as a minister? Yeah, so initially I started as a reverse mentor, a young person under 35 that advised a cabinet minister in 2015. So at that year, for example, Uber came to Taiwan literally without a license. So people with no professional driver's license could start driving strangers for money.

4:45Audrey Tang: and taxi drivers was not extremely happy about it. So they almost occupied the Ministry of Transportation, certainly the streets around it. And so we simply set up a website and replicated the same process, namely people proposing idea statements like insurance or registration or not undercutting existing meters, but search pricing is fine and so on. But crucially, there's no reply button. So there's no way for trolls to grow. Instead, you see your position among people who feel similar to you. You see the other side's position And then the leaderboard is the people who can cross the largest ideological divide, what's called the bridging bonus.

5:20Audrey Tang: You can earn bridging bonus by proposing ideas that your side and the other side can also agree. And so again, after three weeks of rapid convergence, we end up with a set of nine measures that then pass into law. And so the Uber driver, taxi drivers, local co-ops, and so on is a win-win-win situation. And the problem was almost immediately solved online. Of course, it took a couple of years for it to really become law. But it really showed people, actually, we do not need to put everything into binary, into, say, referendum. People can really, in tables of 10, convince nine other people and bubble up their ideas.

5:53So, I mean, you have these discussions, and then the mechanism is that this will then be enacted into law once you pass these thresholds and so on. Can you give us some other examples? I know that in Taiwan, for example, there was a very big problem with deepfakes and disinformation, as there are in many countries.

6:09Audrey Tang: And it's interesting you talk about slop because it is kind of the first time that slop really hit us really badly a couple of years ago. So in 2024, if you open up social media in Taiwan, Facebook or YouTube, you tend to see a celebrity, say Jensen Huang, the NVIDIA CEO of Taiwanese, trying to sell you some cryptocurrency or investment advice or something. And if you click, Jensen talks to you. Sounds very convincing. It's actually his voice. But of course, it's not actually his voice. It's just a deepfake, slopped version of his voice. And so what we're looking at is an algorithmic, kind of perverse incentive.

6:45Audrey Tang: Because it turns out, Facebook's ad algorithm earns more when they push a scam out to you. But Taiwan is the most free in terms of internet freedom in all of Asia. And so we simply cannot do censorship. That is not part of our policy option. So as Minister of Digital Affairs, I sent 200 ,000 text messages to random numbers around Taiwan. like literally random numbers, asking what should we do together. And then we got thousands and thousands of ideas, and then we chose randomly 447 people, a microcosm of the Taiwanese population, again in tables of 10. So each person looking at nine other people on their Zoom screen, and basically they have to think of some bridging ideas that the other nine people can live with.

7:27Audrey Tang: So for example, one table says, let's just label all online ads as probably scan until, say, Jensen Huang or somebody else digitally signs off on it. Very good idea. Another table says somebody lost$7 million to that scam ad. Facebook should be liable for the$7 million. Another good idea. Another table says if they don't want to comply with the liability, maybe for every day they ignore us. We slow down their connection to their video by 1%. Another good idea. And we get more than 85 % people's adoption in all those different measures. They think they are very coherent. and the other 15 % of people can't at least live with it.

8:02Audrey Tang: And so we proposed that to the parliament. And by the time parliament deliberated this law, there's no single party with majority in the parliament. There's three major parties. But because we can show that this is really common knowledge, like I know, that you know, that everybody knows, that these are the measures Taiwanese people consider proportionate. It got passed in just two months. And so throughout last year, there's almost no deepfake scams anymore. Like by now, I think it's down to 1 % of what it was two years ago. So obviously, we can upgrade democracy as a social technology using pretty much the same technology as those technologies that cause harm if we see democracy itself as a social technology to upgrade.

8:41Lots of points we're going to come back to in all of that. But it's great to hear this kind of hopeful example. I think it's fair to say, Matthew, that Britain doesn't quite work like that at the moment, does it? you've been a top Mandarin for many decades, and you've come to the view that really governance in Britain is not fit for purpose. Can you tell us why you think that and what we need to do to fix it? I can, and I do think that our governance system is not fit for purpose. I think in a democracy, our politicians are almost inevitably driven to short-term thinking. We have a brilliant civil service, but it has to do what the government of the day wants it to do.

9:20And so the absolute maximum length of time anyone can look ahead is five years or ten years. And that's the maximum. In practice, it's more like five weeks or ten weeks, and sometimes, frankly, five minutes or ten minutes. And that is the reality of our government today. Governments of whatever political color cannot do long term. They also cannot prioritize, and they cannot join up across the silos. So those are the things that it finds most difficult to do. And at the same time, I think society thinks in general that our governments are there to fix all of the problems. And every time I open my FT, I see the scale of those problems that our country is facing.

9:59Things like the future of technology, the future of the planet itself, the geopolitics, where we'll get our energy from, an aging population. These are really, really difficult choices. And the people must have lots of views on those questions, but no way of actually engaging on those questions. So we co-founded something called the National Strategy Project, which is designed to help whoever is in power come up with a long-term national strategy. And we were really inspired by Audrey's example of how deliberation can give people a voice and can create a sense of positive movement. So I do feel the current system is not fit for purpose.

10:41And I do think there needs to be a bit of a shock, a bit of a jolt to people who think actually everything is just about okay. It isn't okay. And if we do nothing, it will get worse. But there is some good news. I mean, there is hope. There is reason to be positive here. And I do think this could be an example of technology and specifically AI for good. Because we do now, possibly for the first time in human history, have a technology that allows literally anyone who wants to to have a say and to have that say with their colleagues and to create a sense of where is there a common interest? Where does the public as a whole stand on issues?

11:20This is not a focus group or an opinion poll or a think tank. This is about educating people, giving them some facts and figures, some actual specific choices, some realities, some trade-offs, and asking them to deliberate, so discuss with their peers and come up with views, some of which there might be consensus on, on others there'll still be a great divide. But as I understand it from Mordra's research and others, it is possible to find ways to bring people together and to describe where is their informed public consent. All right, I want to get into the mechanics of this a bit more, but I just want to understand, I mean, you are at the heart of government.

12:00You went during your five years in the Home Office, I think you had six ministers who were there. How do you think they would have reacted and how would the civil service react to this kind of participatory democracy where you're getting other people in, you're harvesting their ideas? Would the system allow for that? Absolutely. The civil service is prides itself on being driven by the evidence. You know, it loves nothing better than some data. And here is a way of getting the best possible data into the heart of decision-making. There are already some examples of participatory democracy working across our country and many other countries around the world.

12:38The difference about our project is that this would be being done on a national scale for the first time. And I actually think that's very exciting. All right, we're going to come back to a lot of those issues. But Audrey, you now have this role as a kind of roving cyber ambassador. So you're going around the world seeing how this deliberative democracy works in many countries around the world. Could you give us some interesting examples of who is doing what, where I know that you're in Germany and they have this very interesting kind of concept of the what sucks movement, don't they? So can you tell us about that and some of the other projects you're involved in?

13:09Audrey Tang: Certainly, yeah. I've been in the past couple of years to I think 29 countries now, which is unfortunately the majority number of democracies because we're dwindling, but hopefully we're coming back. But all democracies, as we just listened to, all the civil servants that I talk to are really tired of the peak polarization and the peak slope when it comes to online information. So really, this kind of data is a nugget to them. So in Germany, for example, right now, we're on phase two of the Germany what's bugging you and how to debug conversation. And the what's bugging you part is essentially anyone can complain online.

13:51Audrey Tang: But instead of complaining and just kind of shouting to an empty hall or something, There's an AI, what we call assistive intelligence, that translates what's bugging you into something that is bugging everyone in Germany. So unsurprisingly, the number one cluster is about public transportation. It's about bus and trains and things like that. The other one is about the bureaucratic forms. You know, it should serve the people. People shouldn't serve the forms. And many other very practical stuff. But for the first time, we can do what's called social translation. we can show if we work on this crux, it will then go back to each and every person and say, it's because these five things you said, even those three words you said in the complaint that you put into the What's Bugging You, now this is becoming a collectively defined crux that people can then work on.

14:45Audrey Tang: And so it's very powerful because it closes the loop individually for each and every person for the first time, and it shows them how exactly they can help. And so now there's many, many teams now on what's called quadratic voting, asking people who submit those initial complaints to look at the proposed solutions to those issues and then vote on it. I guess one of the concerns that people have about participatory democracy is that people come up with the wrong answer, however you want to define that, and that people will be horrified by some of the opinions that all of our citizens have. have.

15:18But the way that this works, I think, is fascinating how it worked in Taiwan, for example. So as I understand it, through one of these processes, citizens said that they thought caning ought to... Yeah, caning drunk drivers is a good idea, according to some. And so, but when this process worked out in the consultation, can you talk us through what happened?

15:40Audrey Tang: Yeah, certainly. So basically anything with 5 ,000 digital signatures can force a ministerial response in the online deliberation of some sort. And so you get really wild petitions. There's also one that says we should change our time zone to plus nine, the same as Japan, to force people traveling in from Beijing to change their clock. And that got 8 ,000 supporters. But almost immediately, there's a counter proposal that says we should remain in plus eight and save a bunch of money. And that also got 8 ,000 supporters. And so again, if you allow the trolls to grow, then you go nowhere because people love dunking on each other on such ideological problems.

16:24Audrey Tang: But because again, the online interface doesn't allow replies, but actually you can only propose this bridging statement to earn your bridging bonus. By the time they meet each other in a face-to-face meeting, they tend to coalesce on very useful, what's called the uncommon grounds across both camps. So, for example, on the time zone conversation, people settled on we should do something like the Taiwanese gold card, which is like your global talent, which I'm very grateful to receive. Basically attracting people who work on open source to Taiwan and also maybe hosting human rights conference like RightsCon and so on.

17:00Audrey Tang: That will only spend a fraction of what it would cost to change the time zone. But it will make Taiwan a scene more unique in the world compared to the changing time zone option, which kind of doesn't work. And for the drunk caning issue, people eventually agreed that some years from that deliberation, all the new cars in Taiwan will have to have these detectors for alcohol, so that if you're really, really drunk, the car will just stop starting. Because people, after a conversation with nine other people in the same table, discovered if you're really, really drunk, the idea of caning doesn't really enter your mind.

17:35Audrey Tang: At that point, really the car refusing to start is practical. AI has opened infinite possibilities for customer experience and endless complexity because piloting AI tools is easy. Making them work together across a global enterprise is something else entirely. NICE is the CX AI platform that runs customer experience at scale, uniting agentic agents, human agents, and customer intelligence. So the world's leading brands can move forward with confidence. NICE, the proven CX AI leader. Learn more at NICE.com. For years, credit scores were generated with just a snapshot. A snapshot that left millions of people out of the picture.

18:19Audrey Tang: So VantageScore is modernizing the credit score. VantageScore uses 400 % more credit report and alternative data than legacy models, bringing more people into the credit economy. Like 5 million more mortgage-eligible people. Banks win and francs win. Lenders win and so do lenders. When you build credit scores the right way, everybody wins. Get started at VantageScore.com.

18:45So Matthew, one of the fascinating things that Audrey is describing here is it's a very bad idea to have a polarized debate, which is pretty much the structure of the House of Commons. You've got two people opposing each other. And in this country, we famously had a binary referendum 10 years ago on Brexit. As a thought experiment, what do you think would have happened with Brexit had we had this kind of participatory consensus? consultative process that we went through? I think there is a lot of evidence out there from research that shows that when people are open to this sort of deliberation process, where they have to weigh the pros and cons, that on average people end up understanding the other point of view better than they previously did.

19:27So it's the opposite of polarizing. It does bring people together. That's not to say that everyone will end up agreeing after one 90-minute session. Of course not. But it does mean that people have a better sense of empathy, better able to put themselves and other people's shoes, better able to understand why some people have a totally different point of view from their own. And on average, it does pull people towards some center ground. I think it probably would also show that it's actually not a very good idea to have a very binary debate on something where the country is 50-50 or 55-45 one way or the other.

19:56And I think going through deliberation does create a sense of a stronger ground where most of the people are comfortable. Right. Can we get into more of the mechanics of the National Strategy Project? What are you aiming to do over the next three years? Because I think you want to come up with a big kind of idea before the next election, don't you? Yeah, we want to come up with the first iteration of a national strategy before our next general election. And we want to do that by asking the people what should be in that strategy. And we're going to do that by this process of deliberation that Audrey has talked about.

20:30We have partnership with YouGov and with Stanford University in California, as well as with the University of Cambridge here. So we have the materials that will be ready in terms of giving people that sense of some facts, some figures, some objective realities, some trade-offs. And on what kind of subjects? On all the biggest subjects that you could think of, we will be holding something called the UK in one room, which will be 500 people chosen by YouGov to be representative of the UK. And they will come together for a weekend before the end of the year to do this deliberation on four of the really, really big issues.

21:05Which are? Well, they will be a version. There'll be something about our economy and how to grow our prosperity and where to invest. There'll be something about our democracy, the fact that we have an aging population. There'll be something about the UK's role in the world and our relationships to big countries and how to build up our resilience and our national security. And there'll be one more topic as well to give a sense of the things that really, really matter in terms of the provision of services. to people in this country. But the crucial thing is at the end of that, we're going to ask people, but now you've had a taste of deliberation, what do you think are the issues that matter most to our collective future?

21:44And that bit will then determine the next stage, which is where we'll do a mass online deliberation with Stanford University. And they've got this way of doing that that allows people to come together in small groups of 12. It's all moderated by AI, analyzed by AI, and aggregated by AI, but you can get to hundreds of thousands or millions of people through the platform. And we're going to do that over the next year or two, and then come up with the first iteration of a national strategy, as I say, in time to inform the political parties going into the next election. And just to be clear, we are an apolitical group.

22:23We're not a political party. We don't have an answer to those questions. We just want to ask the right questions and require everyone seeking office, whether that's at national level or local level or anything else, to have an answer to that set of questions that responds to where there is informed public consent. Audrey, you've been based in Oxford for a while and you've been observing this somewhat strange country. How do you think participative democracy could work in Britain?

22:54Audrey Tang: First of all, I think, especially from young people, I hear a lot of aversion of the, And I'm going to quote you just now, this AI listening to us, AI doing synthesis, AI doing summaries. Like people really, there's a real tech clash going on when it comes to AI-mediated conversations. And I think that's very fair. And so a lot of my current work in Oxford is to reimagine AI as assistive intelligence and not as authoritarian intelligence. And part of that is to show that this entire process can run without data centers. It can run just on your laptop, for example. Whatever language models we use are specialized for this particular process.

23:36Audrey Tang: So it's not trying to throw protein and throw laundry or whatever, like everything in the same what's called large language model. Even though it sounds like a technical point, people really want the computer to feel again like personal computing, like you know what's going on there and not being captured or dependent. I think some of the political science literature on this shows that where this does work is when people feel that their voice is being listened to, but that something results from it. Exactly. It's the feedback loop that you have established in Taiwan that something will get enacted if it goes through this process.

24:13Do you think it's possible to introduce that kind of mechanism here?

24:16Audrey Tang: Indeed. Indeed. In Taiwan, just a very quick interjection. There's no guarantee that the government will take on the results because some of these deliberations results shouldn't belong in the government. For example, there's a conversation about the quality of journalism and about the real-time citizen-led cross-checking platform that fact-checks the fact-checkers in the institutional media. Now, that's a good idea if you're in the deliberation space, but it's a very bad idea if the government is to run it. That would be the Minister of Truth or something, right? And so what I'm trying to say is that the UK or Taiwan in one room may agree on things that are bugging us and things that need to be fixed.

24:58Audrey Tang: But it must never be just the government closing the loop. Once it's becoming common knowledge, each person can take it back to their living room. And it becomes Taiwan or UK in five million rooms. And people can discover that this part belongs in journalism. That part belongs in schools and educators and so on. And so everybody can also do their part. Can I just add one thing to that, John? Sorry. So I think it's really important that there is action that flows out of this. I, for one, wouldn't participate in a thing if I didn't think there was a decent chance of something different happening as a result of that participation.

25:30So our project is first of all about the knowledge and then the dialogue and then the strategy, which I've talked about. Those are first three phases. But the fourth phase is crucial. That's the action. So our plan is to set up something we're calling the Institute for Our Nation, which would hold the results of the strategy coming out of the dialogue and make sure that whoever is responsible for whatever it is in the strategy actually does something to take forward the will of the British people. I'm afraid we're very close to being out of time. But Matthew, I just have one last question. You're talking very passionately and strongly about this issue.

26:04Is this the frustration built up over 30 years in the civil service or do you think that the civil servants, this is a pretty widely held view within the civil service? I mean, I love my time in the civil service and I think it's a fantastic institution, brilliantly led. This is not about the civil service doing stuff differently. This is about everyone doing stuff differently. That's why I think it's a challenge to the media. I think it's a challenge to the politicians. It's a challenge to business. Chief execs of companies have got very significant levers that they can pull in all this stuff.

26:34It's a challenge to charities. Think about it, pretty much every sector in this country. And the whole thing could do with a jolt. We tend, in the past, to have our great leaps forward, our big creations of new institutions, only after a war or something really awful like that. It wouldn't be great if we could avoid a war by having this leap forward and creating a different way of doing politics and governance. That's a very positive note on which to end. So thank you very much, Audrey, and...

27:08That was me, John Thornhill, speaking to Audrey Tang and Matthew Rycroft at the FT Weekend Festival. Thanks to them both. Tectonic is produced by Edwin Lane. The executive producer is Manuela Saragossa. Sound design by Breen Turner and Sam Giovinko. The FT's head of audio is Flo Phillips.

27:38Audrey Tang: For years, credit scores were generated with just a snapshot. A snapshot that left millions of people out of the picture. So VantageScore is modernizing the credit score. VantageScore uses 400 % more credit report and alternative data than legacy models, bringing more people into the credit economy. Like 5 million more mortgage-eligible people. Banks win and francs win. Lenders win and so do Lindas. When you build credit scores the right way, everybody wins. Get started at VantageScore.com. Decisions made in Washington can affect your portfolio every day. But what policy changes should investors be watching?

28:16Audrey Tang: Washington Wise is an original podcast from Charles Schwab that unpacks the stories making news in Washington right now and how they may affect your finances and portfolio. Listen at Schwab.com slash Washington Wise.