Introduction

Yeah, damn, I've got a lot to back up there.

But the end of the world thing that you said, well, maybe, maybe we can. Maybe we can do something about that this evening.

So I'd like to ask a question first.

Can you put your hand up if you are worried or anxious about the world at the moment? You know, the wars going on, the divisions in society, the democracy.

Yeah, yeah, that's right. A lot of you, yeah. And there's good reason for it. There is good reason.

But I am going to go through a platform this evening that will hopefully give you a little bit more hope around those very, very topics, the kind of big topics. And we are going to do a real live demo.

And if you want to participate in that, please do. and scan the QR code in order to do that.

And there'll be a couple of questions to start off with before we get into it.

Presenter's Background

And as you're doing that, I'll just tell you a little bit about myself.

So I'm not technical, especially at the moment. I used to be. My first career was in the city designing derivatives trading systems and things like that. Did that for about 12 years.

I got into AI in 1989 in a micro kind of way, but my interest grew with Moore's Law. And by 2005, when Ray Kurzweil's The Singularity Is Near came out, I became obsessed, basically.

Who's read The Singularity Is Near? Has anybody read that?

Okay, so a lot of what we're looking at going on in the world today was predicted back then, and not just predicted, but kind of almost caused because it wasn't just a graph. He said, you know, this is what's going to happen with AI in the next 20, 40 years. But that wasn't just a graph of prediction.

What happened is that the venture capitalists saw that and said, oh, we like curves like that. or let's invest. So it became an investment plan and it became a project plan.

Kevin Kelly said, the editor of Wired said at the time, everybody, every startup's mission statement became very, very clear, take X, add AI, from that book. I read it, I wasn't a startup, but I was still blown away by it because it was so amazing and very, very credible, very, very plausible.

So got into AI kind of from an obsessed amateur perspective about 20 years ago.

The Intersection of AI and Conflict Resolution

During that time as well, I changed career from IT. Yeah, as everybody else is piling into AI, I got out of IT.

And my field for the last 20 years is I've been teaching negotiation skills.

and conflict resolution and i do that across the world worked in about 25 different countries i've worked with organizations like goldman sachs with the british army i've taught hostage negotiators lots of different uh contexts uh i'm a visiting lecturer at imperial uh and i've also taught at the side business school and i've written a few books on the topic

Now, so that's my field, negotiation, conflict resolution, super interested in AI. Obviously, every now and then I've been checking in what's going on in that intersection, AI in that sector.

And for a long time, nothing. And then, yeah, a little bit, but not really especially impressive. And then a little bit more. And then last year I checked in and I was blown away.

And this is Dublin Curse. This is exactly the nature of Dublin Curse.

I was blown away by what was going on in the conflict resolution sector. Really, really amazing stuff.

So I felt compelled to write a new book called The End of Conflict, How AI Will End War and Help Us Get on Better.

less divorces, less arguments with the kids. Well, if we cross our fingers anyway.

Deliberative Platforms and AI

So what we're gonna look at in this session is we're gonna look at a platform called Remesh, which is, has anybody heard of Remesh? No, okay, it's a, okay, yeah.

Deliberative technology. Has anybody heard of deliberative technology or deliberative?

So deliberative democracy is things like citizens' assemblies.

So, for example, in 2016 in Ireland, they tried to address the question of abortion, which was illegal in Ireland. It was in the constitution that it was not legal. Now, Ireland, you can imagine, is an exceedingly divisive topic. Politicians just don't go there.

So what they did is they put it to a citizen's assembly. So a citizen's assembly is you take 100 people, say, randomly selected but representing the demographic spread of the population. You take them through a process. It was over several months, meeting once a month, homework in between, expert input, facilitated conversations and discussions. And then at the end of it, they had a vote. Should it be allowed or shouldn't it?

And there was a two-thirds majority that said, yes, we should allow it. And it then went to a referendum, a national referendum. And again, there was a two-thirds majority that said, yes, we should allow it. So they changed the Constitution to allow it.

What was really interesting was not just that it solved such a divisive topic that politicians wouldn't go anywhere near. More than that, it was a very positive process.

So two-thirds voted for it in the Citizens' Assembly, two-thirds voted for it, but a sixth abstained and a sixth voted against it.

Even those who voted against it, Said they supported it So they were going to support it because we they said we agree with the process. This was a very fair process We were heard our opinions were taken into account We even were able to change the nuance the wording of the proposition But we were the minority so but we go along with the majority vote because we were listened to and apparently

Again, you would expect a divisive kind of conversation, an explosive kind of conversation, apparently all kinds of hugs at the end and new best friends. And also everybody, when asked afterwards, they all said they wanted to be much, they found it such a positive experience, they wanted to be much more engaged in civic dialogue moving forward.

The Power and Challenges of Citizens' Assemblies

So citizens' assemblies are really, really powerful ways of healing society, healing the polarisation, of coming up with nuanced policies that the large majority will support.

But they've got problems. Firstly, they're expensive. Secondly, they take a long time. Thirdly, it's difficult getting the representative sample of the population.

But AI platforms can solve all of those problems and can work at scale to tens of thousands on the conversation at the same time. Tens of thousands done pretty much instantly in a two-hour conversation.

I'll give you an example shortly of something working in exactly that kind of way and coming to an agreement with, again, even the minority that vote against it say they support it because they agreed with the process.

So hopefully on your screens, if you've scanned the code, you will have something like this. Does everybody have that? Perhaps some people haven't. Does anybody not have that, who's tried to?

Did you try? Oh, all right, okay. Okay, cool. So hopefully you'll have, you would have had this and maybe you've answered the questions already. I'll just go through them.

So the first couple of questions are just kind of, oh, not this. Just, I'm not gonna consent to cookies. So do you work in tech? I'm going to put kind of. And how optimistic are you about our AI future for humanity? I'm going to put somewhat. It does depend on the day. Mondays, Wednesdays, and Fridays, yeah, bring it on. Tuesdays, Thursdays, and Saturdays, oh, my God.

So that's just demographic stuff. That's just kind of, you know, those questions, they could be things like age, gender, religion, this kind of stuff. It could be anything. Just to get a sense of who are voting, who's involved in this conversation kind of stuff.

And, oops. Sorry. That's.

And if we go to the moderation page, then the moderation screen, then you'll see already, you can see we've got 25 people who have logged on. They've got 25 simulated responses as part of this as well. We can see the number of people who are saying, yes, they work in tech, some who don't, some who kind of... and we can see how people are such like.

Using AI for Consensus Building

Now, if I wanted to, as a moderator, I could now ask various further questions, because if I wanted to clarify a few things, these have been set up in a sequence, but you can always kind of change that order. And so you can, in other words, you can facilitate the conversation. depending on what you're getting back, depending on where the conversation is going. You can also facilitate the conversation, ask specific questions for specific population segments within here. And so again, you can kind of manage the conversation in this way.

What I'm going to do next is I'm going to press send here. And you'll get a question, how do we solve world hunger, on your screen. And it'll be a free text question. So type in anything you like. You'll have a minute and a half to answer that. You don't need to use all minute and a half. You don't necessarily need full theses. And then after that, you will have other people's responses fed to you. And then you can agree with them or you can disagree with them. Or, and then you'll get more of that and you can agree with them or disagree with them.

Alternatively, what might happen is that you'll get somebody else's, or you'll get two other responses and it will ask you, which do you prefer? And you can make that choice. And again, you'll get a few of those. And this is all capturing the data and capturing the kind of belief map in the room, if you like, in a kind of mathematical kind of way.

So I'm gonna press send. And hopefully, You should have a screen. Why didn't that, oops. Oh, I'm not allowed in. Have you got a screen with a question saying? Yeah, brilliant, brilliant, cool.

So for those people who don't have it, there's a free text question saying, how do you solve world hunger? And you type in what you like and then you get other people's responses. You can agree with them, you can disagree with them. You'll get pairs of other people's responses and you can rank them. You can put them in your order of preference.

And whilst people are doing that, We go back to the moderation page. So we're starting to get responses. We haven't got the, we've got 40 responses. Some of these are simulated. We haven't got the agreement figures yet because we'll wait until the end for that.

So, on your own screen you should be able to see the percentage of people who agree with you and you should also be able to click on the prompt and it shows you a summary of many of the other ones that people have, many of the other ideas that people have put there. So if we were then to have further iterations of that question or clarification questions or whatever, you might be able to go, oh, actually, I like that idea. I hadn't included that. I hadn't thought about that. Right, yeah, I'll put that in my next prompt kind of thing, my next answer.

And basically, the moderator... So what tends to happen, either... through the moderator steering the conversation towards consensus, or quite simply through the iterations and the participants. It's almost gamified to find consensus. So either through gamification or through the moderator steering it in that direction, it ends up going towards a consensus.

There's maybe 64% of people support this thing and 58% support a different but overlapping. Well, maybe somebody else might come up and say, if I can come up with a sentence that I think probably the majority of these lot and the majority of these lot will also support. And so we're suddenly into the 70% kind of region or whatever.

Lots of features around this that you can do in terms of, so for example, let's see. food, food, you can imagine, food has come up, oh, greed, let's have a look at greed.

We can't solve world hunger unless we solve greed, at least in this extreme. Once we crack that one, everything else follows easily. We can't, as greed will always play a part in oppressing others.

Right, interesting, yeah. So then as the facilitator, I could see all this, and I might think, hmm, that's a really good point. Let's hone in on that kind of stuff.

all kinds of analysis that you can do. Analyse. You can get all kinds of, look at different segments. And you can work with it with different kind of segments and this kind of stuff.

We've got, let's ignore the simulated and let's just go for the tech people.

So, Very powerful, very interesting.

You might think, yeah, it just seems a little bit like customer insights.

Peace Polling and International Conflict Resolution

How is that going to bring about world peace?

This chap, Colin Owen. Colin Owen is a research fellow at Liverpool University. He does something called peace polling. So he was involved in the Northern Ireland Good Friday Agreement. And his job there was basically...

He would walk the streets with his clipboard, walk down the high street, him and his team. And they would stop people in the middle of their high street shopping and ask them, what do you want to see in the peace treaty?

And bear in mind, by the way, in Northern Ireland, nobody expected any peace there. It was a... horrible situation, decades if not centuries of hatred and enmity, nobody expecting peace.

But anyway, he asked people what did they want to see in a peace agreement. And he would ask them various demographic stuff as well, and then he'd feed all of that information back to the negotiators. So the negotiators would then say, hmm, So the Catholics want this, they'd never accept this, but they'd be willing to accept this. Okay, and the Protestants want this, they'd never accept this, but they'd be willing to accept this. Okay, I think we can see where the agreement is going to be. And they did this issue after issue. And so the negotiators basically were able to come to an agreement that they were very confident when put to a referendum would be supported. And that's exactly what happened.

It turns out the public input, community input, the co-creating is a key part of peace. So contrast that with, for example, the Oslo Accords, which was taking place just a little bit earlier, about the same decade.

They reached a brilliant solution that everybody in the room was happy with. All sides in the room were happy with. But it was done in secret. It was illegal in both communities for any negotiator to talk to the other side. So it was all done in secret in Oslo. And so then when they presented, as a fait accompli, the solution to the communities, both communities said, no, we're not going to support that. And so it died a death, tragically. So the community input is a key part of creating the peace.

Northern Ireland really was not expected to be successful, the Good Friday Agreement. George Mitchell, the Senator who chaired the whole process, it was three years, he said it was pretty much every week he was checking plane timetables flying back to the States because he thought it was gonna break down irrevocably, irretrievably in that week, every week for the three years. And it was only in the very last day that they got it over the line. That's with, through the night, Bill Clinton, everybody phoning all of the different people involved and trying to persuade them to sign it. And they did. It then went to a referendum and it was supported.

Case Studies in Peace Polling

To the extent, and I'm not saying it's perfect, but here we've got, does anybody know who these two people are? Ian Paisley had a little bit of one. Ian Paisley, Martin McGuinness. So Ian Paisley and Martin McGuinness, broadly speaking, Ian Paisley was the head of the Protestant side of things, and Martin McGuinness and Gerry Adams were the head of the Catholic side of things. They detested each other. Bill Clinton said the first five words he heard Ian Paisley say were no, no, no, no, and no. And they detested each other so much that they would, and this is true, they would bomb their own people in order to blame it on the other side. Both sides did this. It was a really horrific place.

They became known as the Chuckle Brothers. as you can see in this picture, because they shared an office. The deputy, Ian Paisley was the first minister, Martin McGuinness was the deputy first minister, they shared an office, they became friends. Their jollity in the press conferences was so obvious, they became known as the Chappel brothers. So tremendously successful.

Fantastic story, doesn't include AI.

Colin Irwin carries on doing exactly the same, but now using AI, using Remesh, what we've just seen. He's worked in many, many different places around the world, in Libya, in the Yemen, in Sri Lanka, many other places.

I want to tell you about the Libya context. So a few years back, after the civil war in Libya, after it had kind of died down a little bit, but you can imagine pretty chaotic there. All kinds of warlords, gangs, tribal stuff, all armed to the teeth, all hating each other, all wanting to get hold of the resources and such like. And they tried to form a government of national unity.

Now, again, nobody expected it to happen. Nobody expected it to happen in the slightest. But what they did, Irwin ran a remesh conversation with 1,000 people, randomly selected, live on television, on national television. It took two hours.

A third of the country watched it. So they all felt part of it. And after two hours, they came to an agreement on what the government of national unity should look like. And it's still in place at the moment.

So very, very powerful stuff, very, very powerful stuff. And it's cheap, it's quick, it's pretty much instant. That was a two-hour live conversation. You can do it asynchronously if you want. And it's easy to facilitate. So it's a really, really powerful solution. Where is it useful?

Applications and Future Potential

Well, ending war like we saw in Libya. Solving the democracy problem in terms of, for example, the abortion question in Ireland.

Local community issues. Cycle lanes, should there be more cycle lanes, should there be less cycle lanes? Typical thing would be normally that kind of conversation would be exceedingly divisive.

One thing that is interesting actually, if we did this on social media, if I said, for example, there should be more cycle lanes, instantly 10 million people will jump on me, I'll burst into tears and never go back to that question. But here you couldn't do that. Those people who did participate, you'll notice that you couldn't actually do that. You couldn't jump at it.

You could disagree with it. Of course you can. And you can come up with your own proposition, but there isn't a thread. And there isn't a kind of a thread that gets more and more inflammatory as it goes down.

So this turns out to be one of the key parts, one of the key features that enables the conversation to be constructive rather than ad hominem and all of the other faults that it usually goes down. so it can solve local community issues, healing the polarised societies, infrastructure projects.

HS2 or the East-West line, the East-West rail line. These are always billions of pounds over budget, years overdue. Why? Because they can never agree on which is the route.

I don't want it going through my garden, put it through their garden. And they say, well, I don't want it going through my garden or whatever. And so how on earth do we resolve all of this? Well, this is a process that can resolve those.

So we can find efficiencies and not just efficiencies in the infrastructure project, but finding the right solution, importantly. Staff-run organizations, so cooperatives and things like that.

Okay, one more point. Staff-run organizations get the input from the employees. What should our strategy be? Well, if all the employees can come up with that strategy, they're going to own that.

They're now going to be proud to work for that organization. They're going to want to work. They're not going to be slackers staying at home or whatever. They're going to want to work because they're going to love it, and they're going to work so well and so hard.

Aligning AI with Human Will

And the big thing of aligning AI.

How do we align AI? Well, it's really easy. You work out the will of humanity and then you program it to fit the will of humanity. It's really easy.

Okay, how do we work out the will of humanity as the first step? That's a difficult one. There's 8 billion people and they're all different.

Well, these platforms can't quite deal with 8 billion yet, but maybe, just maybe, they will scale up as the doubling curve continues.

Thank you very much.