Alright, maybe a very quick round of introduction just so that we know… Okay, maybe starting with you?

Hi, everyone. I’m Değer. I’ve met some of you before. I’m with AI Objective Institute and with Metaculus. It’s my second time in this room, so it’s good to be back.

Nice. I’m Shu. It’s my first time to be on this side of the table visiting Audrey as a guest. So excited. I’m with vTaiwan community and I think recently we’re thinking about collaborating with AI, that’s why we’re here. And also, Audrey, we ran a catch-up session together during my visit in Taiwan, and I just thought we’ll use this time to invite you guys for a conversation.

Okay.

I’m Colleen McKenzie and also with the AI Objective Institute. And most of my focus is on Talk to the City, the project we’ve been collaborating with moda on.

Here comes a pleasant surprise. I’m also from AOI. I’m Brittney Gallagher, the co-founder. In fact, Audrey, your work really inspired Peter Eckersley and I originally. So, it’s an honor to meet you in person.

Hi, my name is Wendy. I’m currently working in the minister’s office and I’m also responsible for citizen assembly.

Hi, I am Yue-Yin, section chief in the Democracy Network department of the moda. We worked on some CIP projects last year.

My name is Ron. I was from the Democracy Network department. And I basically handle the process of the grant. Thank you.

So, I’m Audrey. I’m still currently moda minister, just the first digital minister of Taiwan. I’m myself transitioning to a more public and intellectual role. I’m also looking forward to work in the civil society with all of you after I transition on May 20th.

Welcome to vTaiwan then.

I never left. I was always in the channel.

Yeah, so this conversation is on the record, but it will not be published on the moda official website. We will make a transcript. We will co-edit it. We will still contribute to GitHub and all that, but only after everybody is happy with the edits. And so, feel free to just take out things that you don’t want to be published afterwards. You’ve been here before.

Yeah, yeah.

Right, so, how shall we start? We did receive a kind of planning document. I don’t know…

Yeah, maybe I can start just to see kind of hing ow the…

I’ll just say something I said yesterday in the g0v summit. I think from vTaiwan’s perspective… So vTaiwan is a community-based project out of the g0v community and it has been host digital regulatory reform conversations from 2014 to 2019. It was more active back in those years when there were more governmental engagements in the process.

And no pandemic.

Yeah, and then pandemic happened. But nevertheless, I think there’s still some people in the community trying to keep the operation just at bare minimum level. So, keep it running somehow. Although during pandemic it’s quite difficult.

And then, recently, we started to realize the energy of the community started to come back. And we do have more energy to continue doing something for vTaiwan. And we realize it’s really important to get a diverse range of opinions, including from industries, from societies, also from the governments of Taiwan specifically.

So, that’s also why we’re thinking from the past year, we kind of start… we revitalize the community a bit. And it’s been kind of running more stably. We have a small meeting every week, meeting every week to basically think about how to host another conversation again. A multi-stakeholder conversation that we can probably bring in new concerns from society and host hybrid workshops to deliberate potential policy recommendations, or regulatory change together with Taiwanese government.

So, it has been like some kind of intention from the vTaiwan community. And it’s also why we’ve been thinking about what kind of topic would be suitable to propose to different ministries in Taiwan. So, for example, we can reach out to ministry of digital affairs and pick a few topics that we might all be interested in talking about.

And that’s also a popular topic. Like the technology we use and the process we use is also quite open for discussion. And that’s also why we reached out to Talk to the City and thought they are a very brilliant platform that we can bring into our digital tools pool and see if that can be suitable for future projects.

So, I guess, we then start to think about, propose this idea of forming a three-way collaboration between moda, vTaiwan and Talk to the City. And as long as everyone is interested in a certain topic that we can host together, from the topic side, from the community side and from the digital supported tool side, that we can, on the process, play, how do you say that, interactively shape a better qualitative content for the topic and a more, kind of, stronger community vibe and also a better, suitable, customized, maybe, tool for the deliberative process for vTaiwan. and the Taiwan society and the ministry as well.

So, yeah, I guess that’s kind of, at least my intention, like, from the vTaiwan community. And, yeah, I guess, if you want to add to it.

Yeah, I can speak a little to AOI’s intentions too. So, we have this tool, Talk to the City, that we’ve built that is in use. It is a discourse analysis tool. So, it can summarize large volumes of conversation, which makes it great for soliciting input from large constituencies, large groups of people. But we haven’t yet built it into something that has the recursive loop, as the recursive public team put it, in the work that you guys did last year, that helps an ongoing conversation evolve and document that conversation.

So, we think that building that in the context of an application to a particular group of people or a set of discourse topics, is the best way to iterate and figure out what the best tools are going to look like. And this is, like, maybe the most exciting application of Talk to the City so far. We’re working with a number of different partners, but the degree to which the government of Taiwan has been interested in the results of our work is really exciting.

And so, I think this is a perfect place to do that experimentation and develop tools that work really well for this use case, because I think it’s a good test bed for figuring out what will work in other places, too.

Great. So, the other community that you work with, I’ve read the Michigan case study, that looks like it also has a chance to grow into a recursive public if you do iterated conversations. But is there an appetite for that in Michigan or in similar communities?

I can speak to that specifically. Yes, there’s actually an appetite to expand it more broadly, because the group that we worked with, Silent Cry, isn’t just based in Michigan.

Exactly.

So, they have contacts throughout municipalities in the states, in Ohio and New York, as well, and want to try to figure out how we can continue the conversation in those areas. And the output of that report, for those who aren’t aware, the TLDR is we interviewed formerly incarcerated individuals in Michigan about their experience returning back to society after having been incarcerated for 20 years. And they are a very under-serviced community in the United States. They’re most overlooked. And many of them have lots of challenges, just like finding housing. The access to technology was a really interesting one. One story was a man was picked up by, returning back to society, his mom picked him up, they drove by a bus stop, he saw a guy who looked like he was talking to himself.

And he’s like, Mom, that guy’s crazy. She’s like, No, he’s on his Bluetooth headset. And he’d been in prison so long, the concept of Bluetooth never occurred to him. And this is something none of us can even think about, because we haven’t been pulled out of society and completely in another place.

And so, hearing their stories, we’re hoping that the legislature in Michigan, in particular, will try to put laws and put forward programs that help make their reintegration easier. Especially on technology, because that’s a really easy one. It’s to start, you know, they know when they’re returning, to start giving them some access to the internet, explaining how things work.

So that’s, in that project in particular, that was just in a small area in Michigan. But we’d like to do it in other places. It was just… it’s a lot of work. I interviewed them all personally.

Oh, you did?

Yeah, I have a background in journalism. And so, I interviewed them all. And so, I knew their stories. It was a very interesting lived experience as the journalist, then see the output, to like, because you know everything you heard. And to see how Talk to the City came up with, you know, the report that it came up with. And the, it was a very, very interesting experience.

But yes, to the point, they want to do it in different cities, to be able to take that, you know, collective artifact and share it with, you know, leaders in the legislature there. But I think to continue, you know, to continue the conversation is something they would also be really excited about. It’s just funding and figuring out how to do it.

So, there’s a third case that I want to mention briefly. We don’t have a public output for this, which is with labor unions. Labor unions only have in the U.S. a 20-day window to negotiate with the government on changing their salary and benefits, which is crazy. Like the system is rigged so that there is no room for deliberation.

So, we worked with the president of the California Chapter 4 VA, which is the Veterans Association. He sent out a survey. We collected the results, consolidated it. He sent out the report to everyone and asked for feedback. And just with what exists, we were able to do a loop this way where like no one is expecting a 24-hour turnaround on a report like this. And most people just don’t even submit anything because they are quite checked out from the union work. But when he was able to get back to them with like, here is the report round two for feedback, the engagement I think went from 200 to like 4,000… I don’t remember the numbers, but because when you see other people start saying things, that’s when you say more.

So, I think the appetite would be quite, quite powerful, especially if you have a short one.

Was it also with the video excerpts?

This was purely on text. But he… I think he has identities of everyone, so he can follow up with them. The problem with VA was that not everyone speaks English. There was a lot of Filipino. It was a multi-language piece, so there’s translations that are necessary. But I guess it could be video?

That was, I believe, one of the use cases where we were talking about audio being a possible input because then we can get transcripts from that, type it into Talk to the City easily, but it’s much easier for people to just speak fluently in their native language and then for us to translate them.

Yeah, I think if you look at Cortico, that’s their main mode now, right? Their whole webpage has been redesigned so that it’s audio excerpts first, right? And making that first class instead of clusters first class, I think, speaks to kind of this lived experience, how people feel about having a conversation about this sort of thing. And people feel heard, like literally heard.

So, I think that’s very powerful. That really is very powerful. Because in Taiwan we have many governmental agencies that regularly run deliberative conversations that is set by the citizens or at least the stakeholders, but they’re all just tallied on the same website through all the very different organizers. But almost always just in summary form, or summary form plus a, I don’t know, YouTube video, live stream, but nobody has the time to, you know, review them all, right?

One very good example is called Let’s Talk. It’s at youthhub.tw. It’s just called Let’s Talk, right? So, very year there’s dozens and dozens…

Can you copy it into the document?

Yeah sure, I’ll just start dropping random URLs here. Okay, great. So, yeah. as I was saying, just this year alone, there’s going to be 34 different CSOs, each hosting at least one deliberative workshop on a shared topic, which is called housing justice, and which is one of the main issues that Taiwan, not just young people care about intergenerational as well, but the young people feel strongly about it and which is why they volunteer to deliberate about it and why the youth development agency sponsored this work. And it’s part of the open government partnership commitments for us to do things like this.

And so… But if you look at kind of previous years, for example, on net zero, mental health and so on, these were some of the previous years’ topics, you do see some of the metadata. You do see the handbook, but you don’t see much else at most AI live streams. So, in a sense, they’re heard only on that year, in that very local level, and maybe became like five words in the summary. But the policymakers, if we want to go back, there’s no easy way for us to go back.

And this is, I think, something that they… Like, this is what we call a cheap and cheerful win, right? Because you don’t have to do anything extra. All of these are strictly a subset of what you did already do with us, with Michigan, and with the labor union people. And so, that’s one obvious entry.

So, the Youth Development Agency is formally in charge of spreading deliberative democracy in Taiwan. And the person who established this whole program will soon be our Deputy Premier, Cheng Li-chun, previously Minister of Culture, previously Minister of Youth Engagement. And so, I think this is like the obvious natural ally.

And I think vTaiwan has plenty of people who feel strongly about housing justice as well in adjacent communities. There’s this Rentea group, there’s many projects within g0v that also has housing justice as their main concern. And they were just, I think, drawing this impossibility event diagram. Like they can’t both advance social cause and mobilize in an open way and being government funded. That doesn’t work, right?

But I think the government can play a catalyzing, a convening role to put together the democratic innovators, that’s you on this side, and the CSOs on this side through something like the Youth Development Agency. So, this is just one of the first ideas that popped into my mind once I read your proposal.

Yeah. That makes a lot of sense. Was I hearing in what you were saying an implication that there is also past data that we could perhaps summarize and make more accessible?

Because I was the minister in charge of youth engagement, so we made sure that they keep plenty of past data. It was just never analyzed, never-mined, so to speak.

Maybe we can help with that.

I know. I know.

Is it audio form? The past data.

Some of them are in audio, some of them are in video, some of them just text. Because it’s both about training upcoming facilitator teams and also writing facilitation, so it depends on their record keeping capacity. But it goes all the way since, I think it goes back to 2013 or something. So, like, at least 10 years’ worth of data. And in recently, I mean, especially during the pandemic, they experimented with video form and that is like, of course, all in recording because it was remote too. But the face to face one, I think there’s also plenty of record keeping there as well. So, there’s like…

Seems like a very great avenue on like already something past exists up until now, what’s the next round?

Right. Exactly. Yeah, there is more like 100 past deliberative workshops worth of data.

It’s worth asking questions, and this is kind of maybe more in the weeds, but how should the opinions depreciate through time? There needs to be some sense of calibration like if you go on and if the minute of your scroll is everything about 10 years ago that also doesn’t seem right?

Unless you’re a historian. Or investigative journalist.

Yeah, anyway, I think these are more in the weeds, but thinking about how to bring the newest things to the top of the stack and seeing how it relates to the past…

These are the kind of questions we would want to be investigating.

Right, so…

Yeah, so I think one immediate action is just for DN to maybe through the cabinet level youth advisor network, which is also run by the same youth development agency, to introduce you to the YDA who runs Let’s Talk.

And I think there will be a lot of investment in this because that’s like one of the main promises of the incoming cabinet is just to reconnect with the youth. And so, this will be like a natural ally for you.

That sounds perfect.

Okay, sounds great.

Yeah, especially if any of the upcoming consultations, whatever number you said were happening this year. 34, yeah, have a bunch of past data that is relevant. This sounds like exactly the kind of situation where we would want to be applying new in person tools.

Yeah. Are there any thoughts from the DN?

I think… So, there is not a specific topic we want to talk about right now.

We were talking about housing justice, which is the YDA topic.

So that’s the topic we’ve chosen?

Well, the YDA chose that for this year’s Let’s Talk already. So, they are like with or without us, they’re going to run 34 workshops. It was just the main value proposition is that so that they can aggregate better, right? So, it’s not like vTaiwan will run two additional workshops, not like that. More like vTaiwan will support those 34 workshops.

I probably just want to use this chance to also like bring this default topic we proposed in the document. And just have a kind of slight comment on them. Because at least initially from the vTaiwan community.

Like, one of my concerns is actually, maybe you can help me out with the process design, because vTaiwan has been very tech heavy, so we always discussed about the topic we choose will be more ideally if we choose topics relate to digital issues.

Like not integration of incarcerated individuals.

But on the other hand, right, because Talk the City has already been involved in like union labors and Heal Michigan, this kind of work. And these were not really digital issues per se. It’s more social topic that still can be discussed through these platforms.

Right, right. They’re more in line with like youth engagement on social justice and housing justice is at the same level with things.

Yeah, so I’m just thinking like is this a moment to modify the vTaiwan principle in a way like the issues we pick doesn’t really have to be digital related issues. And I’m just thinking about why is that, if this change happen, even because the digital layer is actually more easier to use, and is more kind of widespread into every corner of the citizens, and people are more easy to access to digital tools because there was the reason initially why we thought…

Yeah, 10 years ago.

Yeah, 10 years ago. So yeah, I mean, I welcome this change. I just wanna share my concerns as well and see if there’s any feedback, advice or inputs from you.

Yeah, I mean, 10 years ago, the vTaiwan scope was limited precisely because the… Like, of course, Uber was controversial. Marriage equality was more controversial, right? So, if we have the tools, why not to solve it like the actually the most controversial topic?

And the reason why is that not many of them are online, right? The people who are highly religious and was not so pro marriage equality. They don’t feel very comfortable if we say, you know, come to this Taiwan platform or that we’re going to, you know, just interview you, but then check out POLIS for your ideas.

It doesn’t sound right. And so, it didn’t feel right. That was that the main thing. And I think with the advent of technologies such as GPT-4 Turbo, Claude 3 Opus, Llama 3, TAIDE, and so on, now this is being bridged, so that the stakeholder groups no longer need to adapt to technology. The technologists can adapt to them.

Back then, it was almost impossible without a lot of volunteers to consciously make this bridge. We can’t just go and interview a random mini public of people on marriage equality and expect them… their ideas to be accurately represented online. There was no such technology. It would be so expensive. That is, way beyond the vTaiwan community’s means. But now, well, with your help, that cost has been reduced to essentially zero.

And so, I think just this, like, what’s the inverse of broadcasting? Broad listening. Whatever. Okay, broad listening. Yeah, change the nature of this recursive public. The mediators were no longer there. They were no longer one way only. The mediators were just broadcasting, essentially. And it depended on the rest of the facilitators to do the other way, which doesn’t scale. And now this also scales. I think that’s the main difference.

Honestly, what you just said sounds very quotable. I would love to know…

No, actually, I think this is really powerful. Like, the word play of, like, broad listening can actually be scaled. Maybe we need a name for this now. It’s very powerful. And, like, the comparison between 10 years ago versus now being, you know, here’s the shift, so let’s have that reflect now. I really like that.

Thank you. Yeah, one point of example. The AI Evaluation Center, which was under moda’s purview, recently just run an alignment assembly on information integrity. And it forced something, like, between what vTaiwan would traditionally talk about, which is platform governance, but also with everyday people. Because we just sent 200,000 random SMS to the entire public. And they fill in the survey, and we did stratify random sampling to make sure the mini public of 450 people are statistically representative of the Taiwanese public.

And so, it’s like a poll, but it’s also deliberative. It’s facilitated using the Stanford Online Deliberative Poll Platform, which is like a robo-facilitator. And so, the 450 people includes more than 100 media practitioners, like people who are on the first line getting impacted by generative AI insulin. And they get sorted 10 at a time into chat rooms and video calls but there’s no human facilitation. So, it was all on the record, machine facilitated.

And so, we have a lot of raw data, actually, as a result of that, which may also be AI analyzed, actually, now that I think of it. But the thing is, though, that it is as rigorous as a regular poll, but far more agency to the public, because in addition to the existing measures that we want to gauge the social acceptance with, for example, requiring digital signatures on advertisements now because you cannot tell a celebrity from the deep fake and so on, they also set a lot of agenda, like how their ideal PKI system should function in a way that’s decentralized, that’s not state controlled, and so on, which you won’t get from a regular poll, right? It only gets interesting when there’s high bandwidth input from the citizen in the local city, either polis or synchronous like this one.

So, the upshot is that I think we can now safely assume that this mini public has all used something like Facebook and YouTube now, which is not the case 10 years ago, right? So again, this whole of population sampling was something that Vita did not do because self-selection and rolling wave surveys were assumed to reach the people who are actually relevant in public.

But now, the relevant public is like literally everyone because everyone is online now, right? So, then we can afford to do randomized stratified sampling and not get caught in this informed gap of like a lot of people, half of them don’t know what this means. So, now what you do, but now this gap is being closed very quickly.

Did I answer your question?

Yes, totally. I also just answered myself another question I had, but I can just figure this out. Because in the beginning, like within the community, we thought that, you know, we’re going to do this. Within the community, we thought the value of Vita would be doing something that government hasn’t done already.

But just listening to this potential topic for housing justice, I think it’s really, it’s also quite fair to say that if the government is actually hosting some kind of public listening, we can also support from the community side. And support with the kind of creative process and creative digital tools. And see if we can better the process together. So, it’s not like really just kind of, you know… I think that it’s a softer approach and I think it’s great.

Yeah, and also that is partly thanks to Shuyang’s work back in PDIS, right? Because 10 years ago, there was no like generally agreed norm of starting consultations like this and expect it to not break, right? There was a lot of fear, uncertainty and doubt. Like if you start listening at scale, maybe some people get violent. Maybe all we get is noise or things like that.

But then we, together with Wendy, also started a radically transparent experiment so that people can see that such consultations, they always result in some signal. Not all of them actionable, maybe half of them. But it never ends with an explosion that destroys the public trust.

And so, I think that in the last 10 years, there’s now this very relaxed attitude. Like, yeah, of course, it’s just, you know, like a higher quality poll is just from it, so that the appetite for deliberation in Taiwan now is not something avant-garde anymore. It’s something that we just run it as a matter of course.

I would love for that to become more mainstream.

Yeah, I was saying that we’re all glad to be here and experience how that is.

Yeah. How can we accelerate that?

I mean, honestly, one of the first things I think that we should do, like, as a much higher priority this year is actually write a lot more and share a lot more about what we have done. Like, last year when I came here, we did not have the case studies to showcase. It was like a budding idea.

Now, even with these, I think, like, often I would present things at conferences and people say, why have I not heard of this? And I’m like, maybe I should have Twitter, but like, that’s like a… So, I think there is a strong push on that front. I’m personally excited that, you know, your next chapter has a lot more of like a public influence flavor.

Which means I can amplify your voice.

I think the main thing is just making things mainstream. Yeah, cause I would love for this to be a no-brainer that if this is not happening, that’s when someone is suspicious. Then the other way around…

I think what we in the vTaiwan community 10 years ago was doing was a lot of taking the best from Consul, Decidim, Better Reykjavik, from the occupy movement, from Loomio and so on, and try to see whether a combination of these in a toolkit approach, instead of a solution-turnkey approach works locally in Taiwan. And I think that experiment worked very well. We did conclusively prove that a toolkit approach works better than a solutionism approach.

The kind of paradoxical challenge now is that everybody now thinks there’s something about Taiwanese exceptionalism like…

But being here to be fair, I’ve noticed some Taiwanese exceptions…

Please, I’m all ears.

Like the public transit situation is up front like, it’s all just so easy so I really just… Everyone ‘s so nice…

You can take a nap in the park and don’t worry about your belongings.

Yeah, it’s very refreshing.

I was at the jade market yesterday and I was looking at something and she says, give me one second, and she just left. A I’m there for like five minutes and it’ just like all of our jade things are done. I’m like yeah, I see…

Radical trust.

Yeah, radical trust. I love that.

But yeah, the some of them, I would say it’s not just a precondition, it was also the result of the decades work. Yeah, because ten years ago, we were at the cusp of very polarized society. And I mean even before the Sunflower movement, there were a lot of distrust of the government and the state apparatus. And there was a real danger of being polarized the way the US is right now with the administration enjoying 9% of trust from the citizenry.

And if we did get polarized along party lines back then, I don’t think there will be much radical trust as you observe now. And the fact that we didn’t polarize along the party lines back then, and the fact that for the past ten years, a more… I would say, depolarizing norm of democracy has been adopted.

Of course, it’s partly thanks to the trust environment that Taiwan already enjoys but I would say that over the decade, it also contributed to the radical trust that you are observing now.

Actually, I have a question to build on that. So, I guess to both of you maybe this just seeds the idea. When we last chatted here, we discussed a couple of points and we were talking about having an API so that others can use LLMs downstream. Did that actually happen? Like we were talking with Billy who built a whole like line integration and we were like this is great, maybe we should merge it.

We prioritize much more that than say like a single like UI that solves issues. And I think one of the things that as a result we didn’t prioritize was around how to set the scene up for depolarization. And especially given that this is going to be trying to amplify the recursive aspect, I think what we surface as… I think it’s obvious to point at something like it looks like here is some work between these specific ideas but doing that the right way is important because it sets up the stage.

And yeah, just curious to ideate on that a little bit because I think that’s kind of the interface work that’s going to set up whether or not this is bringing people more together.

Yeah, and I think there are like new helps from the ecosystem as well, right? The jigsaw perspectives API was just updated to include bridging attributes, right? So, in a sense, you don’t have to rely on GPT-4 Turbo alone you can also consult the perspectives API I think it’s affinity, compassion, curiosity, nuance, personal story, reasoning and respect. And it used only to do toxicity, hate, whatever, like the negative stuff.

But now, I think jigsaw has been pivoting to actively surface the bridging stuff. And the great thing about having language models that actually parse these attributes is that if you deploy it in a kind of synchronous fashion, it can actually actively steer or nudge conversations.

Imagine like in the Stanford online deliberation, instead of just nudging people who haven’t spoken yet, you give a like reward, a sticker, a badge or something to people who just relates a personal story or adds nuance to the conversation and so on, that’s going to be very powerful, right? So, even in asynchronous mode I would encourage the API remixers to join the AOI APIs as well as the new jigsaw perspective APIs for like zero cost bridge making essentially.

The APIs make it super easy and also being able to almost like cherry pick into like I want to use this, this, this, or I want to write my own prompt and try it with an open-source model. That is all technically possible even now with the graph parts that we’ve prototyped, but how does this come up in the interface I think is a whole department. I mean this is more for you I guess. I mean don’t need to have an answer right now but like… I think this is really the meat of like what I find to be. Like I have not yet seen a UI that makes me intuitively lean in towards bridging. And so yeah, I’m just curious about…

Yeah, I mean this scope of making things recursive even though I’m still learning what it means to be recursive. Because like in the beginning, initially I thought it would be great to consult for a group of people, maybe they had conversation 10 years ago and just consult for them what they would think and simulate how future population would think about the current issue we’re discussing.

But then, when I study more it will be beyond just consulting different 10 snapshots of conversations. We’ll be making I guess the outcome of each conversation recursive as well. We’re not stagnate to a certain result of a conversation as a snapshot, but actually, the outcome will be evolving and you have the potential to track how people change their minds as well.

And all that layers of potential is very, very exciting. Like when you said the difference between… Yeah so, just saying I think it’s part of making things more synchronous. It’s also very interesting because when we thought about asynchronous conversation and having… So, we thought about say for example using POLIS and using Talk to the City, if assuming they’re asynchronous way of engaging people, you instantly open this whole new world of… you can consult for previous conversations because it’s asynchronous, anyway.

But then, if thinking synchronously like Audrey said, you’re empowering the kind of drive of conversation. I guess, like you’re driving conversation in a way that’s actually making that result iterating faster and more abundant, that’s really amazing.

Like there’s this experience I have when I’m on Slack where like I go to bed and I wake up the next day, and the Europe folks already have like a 35-thread deep thing on a discussion. And I’m like I don’t even know where to end it like if we can counter that in the async context, that’s really powerful, right? Because it’s like here’s a 10-year long depth of all the data but also like, you don’t have to go through the rabbit hole is one way…

Also seems like there’s an opportunity for having the best explanations of a particular perspective best arguments or something, like make their way through discourse more and more people can easily reference what happened in the past.

Right, like the context.

Like a democracy of conversation…

Yeah, context is very different in different time frames. Like 10 years ago how people would interpret their results it’s totally different, but the timing we insert language model into interpretation would be very interesting as well.

That’s true, being able to bring that context forward.

Right. Yeah, that’s right.

If you have any ideas on flows or interfaces…

Yeah, please guide us. I want to formally invite you to… mentor us.

We can continue this conversations after May 20th.

Yeah, so I think we have a clear scope and there’s this like making the toolkit approach even wider in a sense like applying AOI not as new conversations but also to existing or historical conversations. There’s this general willingness to try integration with existing APIs and not just perspectives jigsaw but also other g0v projects as well.

For example, I can very easily think about integration with Cofacts, that’s another very powerful… And Cofacts is not just Taiwan, it’s also Thailand so it proves that it’s not exceptionalism.

Actually, I like how in Thailand there are religious communities is what I’m…

Yeah, I think it’s a very interesting avenue of driving into the religious communities of like fact checking. Because if you look at our if you take our country as a case study there, one of the biggest issues with facts is through religious communities and how certain ones are not seen in a specific way. And if you could start with certain groups who are more likely to operate on the same set of facts as most people, you could maybe start to influence the other fringe groups. And I think it’s really interesting.

Yeah. Also, in the US I believe people who are religious are also a bit more DEI diverse, right? So, like the idea of, you know, more inclusivity in the conversations and so on, religiousness is one kind of proxy indication of that.

In contrast, in Taiwan the more educated or scientific trained you are, the more likely you’re religious, so this is like opposite to the US conversation. And mostly because our religions are really soft…

They’re pretty chill.

They’re very pleasant. They make a lot of sense. I’ve been to lots of temples in the last week and I’m like, this is beautiful. I get this, like I love this.

Yeah, so our current premier Chen Jian-ren, John Hopkins-trained epidemiologist did say on formal interpellation in the LY that God does talk to him. So, that was part of the LY proceedings. Like everyone is very chill about his catholic beliefs.

Alright, any more ideas from DN?

I think we’ve talked about that we want to use some g0v project to go through Talk to the City and Cofacts, right?

Yeah, if it’s possible. But you already know Cofacts people, right?

Yeah, we already talked to Johnson, right?

Yeah, we will follow up and figure out the right scope.

Awesome, awesome. Yeah, and there’s also our existing alignment assembly from the Stanford deliberation in the past ten years of the youth hub, so maybe what we can do is just introduce AI and become community to the use of people.

That’d be great. One question about the Let’s Talk programme, each year it’s a different topic, right?

Yes.

So, there’s not there hasn’t been a topic on housing disparity before?

Maybe in smaller scope but not the yearly theme. The yearly theme is co-created by the youth like every year, so they have different concerns every year obviously.

Honestly, I think that I can easily see is if we have a focus on housing and have interesting data sets trying to find a mirroring project with Rose and David from Plurality Institute because housing has been one of their main…

It could be interesting to see you know rhyming issues yeah across very different regions as you know different dockets.

Yeah, and it brings it closer to the mayors, because like information integrity — although a very good topic — it’s almost always a central government problem. And so, it has to be abstract and when it’s actionable, it’s usually on the federal or central government level. Of course, California is different because it houses many social media companies.

But housing, you know, any small towns mayor can deliberate about housing. And so, it connects to potentially to more CSOs because it’s not just those CSOs in Washington DC doing AI governance work but rather the CSOs in each and every city.

Okay, good. Well, thank you for your time.