Got it. Great. Cool. I just think it’s it’s helpful because when I was diving into your, Audrey, the nuance and how you create spaces in which conversation happens, I think is is actually critical and deeply thought through.

Yeah, I see. And so the critique that vTaiwan hasn’t been used for that many major decisions since 2018, is referring to its previous like self selecting people opting into going there on their own, versus the new systems that are doing the mandatory polling, and the the five thousand ...

And then there’s also you were mentioning earlier that text message service where it will randomly text some selection of people when they get texted, did they land in vTaiwan or do they land in some other process?

Just to make that one notch simpler, so the Join platform is a petition platform in which if five thousand people are demanding the same thing, they get the attention of the government to get facilitated and to be a part of a deliberate poll.

Because he says since the government is not mandated to adopt recommendations coming from vTaiwan, legislators don’t take it seriously. What is your response to to that critique?

There is another critique. And I was wondering, Audrey, if you if you wanted to steel the critiques that have been made about vTaiwan. I know that there was a former Taiwanese legislator, Jason Hsu, who said the platform, you know, although I think there’s many rewards of it and ...

So we might want to do the critique, though, question about you brought up one critique, Sasha, that how do you get a requisite variety of citizens? Isn’t it just the digital ones? And we got an answer from Audrey on that.

So before we do that, I just want to make sure if we do have more time, so are we all here for at least? Because I don’t know if you were ready for that, but we can all do another. I can do half an hour to an hour. ...

Want me to tee that up, Sasha? I’ll just say that. So, Audrey, how do you get over the selection bias effects that there’s going to be certain kinds of users, maybe more Internet centric, digital centric, digital native users who are going to use this, but then it leaves ...

There’s so many things in there…

Because ideally I would like people to point this episode, shared around in every democracy in the world is like, how do I get the Audrey in for my democracy I should do these 20 things. And we’re giving people a hint of it.

So, I know we do want to talk about why Taiwan is a target of all of this in the first place, and some interesting facts about Taiwan’s relevance in the world stage, including the fact that it manufactures the world chips for AI. We could do that, or I ...

Do we need more explanation on this? I just want to make sure before we move on, because a lot of people don’t know.

And that’s a theme throughout your work that I want people to see, is that these are general strategies for how do you make sure the immune system is stronger than these supercharged viruses, and you’re presenting some principles for that.

And the second thing I heard you say is redundancy, that if one system fails you’re designing of two other backup systems that will automatically switch over to. So, resilience, it’s a principle of variety and diversity, so that your best defense that creates that resilience is having multiple backup ...

And then, how do I defend against all of them, so it’s like, what are all the viruses people could make and then let me pre engineer all the vaccines for all those potential viruses, so that I’ve basically defended against them.

So this is so important I want to make sure listeners get this because what you’re also saying here is just like you’re pre-bunking, the kind of information threats that come in; I think you should explain Pentera, you’re basically pre-calculating what are all the computer attacks that they can ...

Another place to go from here is, you know, in a conversation we’ve had previously, you talked about like what’s required on the cybersecurity side, and I think you were talking about, you have to assume that your systems have already been hacked and so I just love for you ...

What you’re providing here is an answer, or a proto-blueprint for the response that democracies could have, that left them off stronger rather than weaker to AI. And that’s why we’re doing this episode — that’s why we wanted you on. This is just really great.

You’re speaking to something that we’re trying to get ahead of for other democracies, because here we are in the United States, what’s to stop China or Russia from doing the simultaneous denial of service attack, hit the colonial pipelines, hit gas stations, ATMs, banks, overwhelmingly combined deep faking information ...

That’s an amazing example, I think that’s an incredible example. First, it’s incredible work that you’ve done, Audrey, that you have crafted these responses to what could have been a real panic-inducing attack.

Talking about, and then her first answer was about how do you use large language models to strengthen the immune system, so it’s not that large language models are the virus. It’s the immune system can use larger models to strengthen itself.

Yeah, that was the premise is that what large language models do to democracies, and then we enumerated this.

I didn’t quite understand the threads poetry thing, I just missed that.

But it’s not just a solution to that, it’s a synergistic satisfier, it does more than one thing at the same time. It also brings people together in this deliberate polling, actually puts people face to face into community to work through problems, and so we get sort of a ...

I think you call it the bandwidth problem, or the bit rate problem, that right now humans just give one bit of information every four years to decide which way the country goes, and we could be doing it at a much higher bandwidth or bit rate.

And the same thing is the political philosopher Hannah Arendt’s point that totalitarianism stems fundamentally from loneliness, and so what I’m hearing you saying is that there is a solution not just to better voting.

It reminds me of the author Johann Hari’s line — “The opposite of addiction isn’t sobriety, it’s connection.”

We’re trying to fight both the loneliness national security threat, and fight the individualized sort of outrage-bait information, with this bridging information.

But who agree with them about this more bridging consensus, like we all agree about this bridging statement. These are the things we all agree on. So when people feel membership or belonging in that thing that I care about that is the bridging sentiment and I’m bringing people together, ...

But then, what you’re saying is, well, what if we group people together into communities where they feel this deep care and this deep reflection of their values, which is what your vTaiwan and deliberate polling type systems and structures do, is they invite people into seeing the people who ...

So just to make sure I think for listeners to get this, so it’s like, okay, so people are vulnerable to a one-on-one attack of one person, creating a fake relationship to influence another.

There’s just a lot in it, and I think we should just explain it a little bit about maybe. And, Aza, I feel like you have the best expertise to try to do this one, if you want to try.

I feel like this is a good opportunity for you to jump in, I feel like we should. This is great, Audrey, I think we just need to like explain it well, for each of these things.

And so we might have previously live in a world where, you know, most of the world’s attention in your country is like on a handful of channels, and a handful of stories are going viral. But now, we can start to do a much bigger set of diverse things ...

Yeah. So, I guess I have a question about that. I mean, isn’t it possible that AI will enable a much more diverse set of future deep fake categories and channels, so I could say what are the 10 memetic tribes are, like political tribes that I want to attack, ...

What is it called? Crowdsourcing of facts and responses, and notice of those things, you’re basically expanding the collective compute of the immune system, to try to match the collective compute of how much the virus is evolving, is that right?

And then I heard you also say, just like open source AI development, is almost like expanding the number of people who are expanding the virus, but you’re also saying, I’m going to do open source collective crowd sensing.

Because if you can train it so instead of having to pay a human being to do a lot of work, human-manual and human-cognitive labor, to generate what should be in the fact check, we can make it generate it really quickly.

What if, instead of saying that’s just the virus in my previous analogy, what if we said that virus we can, the immune system can use the fact that we built this virus, for strengthening the immune system. So we’re actually channeling the better and better language models and AI ...

So I heard you say a few things. One is that, okay, we have hundreds of billions of dollars in venture capital training these, you know, large language models and AI systems.

Okay, so this is great. Let’s make sure we’re breaking this down for listeners so that everyone’s understanding.

One, you’re sort of teaching all the adversaries, like how to actually make deep fakes, aren’t you accelerating the problem…

And before we jump into there, I’m sorry because it’s like such a good question is, I could imagine people in our audience, listening to the example you gave Audrey of, like alright, we need to pre-bunk by showing people how deep fakes work, and I could imagine the audience ...

So I think the principle that you’re going to be speaking to, throughout these answers, Audrey, is how do we make the immune system computation stronger than the computation of the virus evolution.

And democracies or governance of societies are like the immune system, and they don’t have hundreds of billions of dollars of venture capital and open source of, you know, governance hacking to make your immune system conventionally stronger.

So think about you have a virus, and you have an immune system, and the amount of compute power going into the evolution is AI, which is generating all these new kinds of threats and mutating every day. There’s like billions of dollars of venture capital, open source, collective intelligence ...

Maybe, maybe just to ground, let’s move to the other one. Maybe just to ground it, I think the principle that right now, the compute power, going into AI evolution.

Then I also heard you say, to give an example of what people might do to attack trust, and then develop antibodies.

That’s a great, great level of example. So I love that example, because you’re pointing to both the need to understand the threat — in cybersecurity you let the defenders know ahead of time, so that you build up the antibodies, you try to patch your system, before the attack ...