Latest / Rewire To Rise– Beyond The Mindset / AI & The Human Future | Episode 3: AI & Democracy (Live Panel)
Transcript
- 0:02What happens to democracy when every citizen no longer meets
- 0:05the same public world? When information becomes
- 0:09personal? When reality becomes filtered,
- 0:12When trust is shaped by the platform we use before we even
- 0:15notice it? AI does not only change
- 0:19technology, it changes attention.
- 0:21It changes doubt. It changes what feels true.
- 0:25It changes the human state from which we participate in
- 0:27democracy. The question is no longer only
- 0:31can AI answer us. The question is, how does AI
- 0:35Orient us? Does it help us think more
- 0:38clearly? Does it help us stay human under
- 0:40pressure? Or does it divide us into
- 0:42private realities, each of us guided by invisible values,
- 0:46hidden assumptions and models shaped by cultures we may not
- 0:50even understand? Democracy needs more than
- 0:54intelligence. It needs shared reality, It
- 0:58needs trust. It needs humans who can still
- 1:01pause, doubt, listen and respond.
- 1:05Join us for AI and the Human Future episode 3 AI and
- 1:11Democracy May 31st 15 wound CET live on YouTube.
- 1:18Because the future of democracy may depend on the state of the
- 1:21human being inside it. Hello everyone.
- 1:33Let's wait a couple of minutes for others to join as well
- 1:38together with us. Thank you.
- 1:44So just a couple of minutes. OK, so welcome to life event #3
- 2:02AI and democracy. Today we are moving past the
- 2:05surface level hubs to confront a massive hidden tax, the
- 2:10unqualified cognitive death of living and leading in a world
- 2:13running on automated systems. Well, with China moving at
- 2:18terminal velocity, Europe cannot afford to lag behind.
- 2:22AI technology and data will entirely dictate the future of
- 2:26our trade jobs. Before we start the live event,
- 2:30Please remember to support the channel by subscribing to Rewire
- 2:34to Rise. We truly cannot move forward
- 2:37with this mission without your support.
- 2:40You can also scan AQR code that you're going to see on the
- 2:44screen to sign up, ensuring to receive the exclusive event
- 2:48Goodie bag from our specialist and me directly in your inbox.
- 2:54And with all being said, let's get started and take the 1st
- 2:59guest speaker which is used Hello, used welcome, hello.
- 3:05Marella good to be here again. Good to have you here.
- 3:11The floor, it's all yours. Excellent.
- 3:14Well, thank you Marella for that.
- 3:19And I would like to kick off today's session on AI in
- 3:25democracy. For those of you who were not
- 3:27here the last time, I'm a associate professor with the
- 3:32Business Academy Copenhagen, where I teach e-commerce and
- 3:37other good stuff. And where I'd like to to start
- 3:43off with here is well, before we we blame the machine, we need to
- 3:51know what we mean, specifically what we mean when we say
- 3:55democracy, which is the topic of today.
- 3:58So I'm going to to start out by talking about what I perceive as
- 4:02some common misconceptions about democracy.
- 4:06Then I'll introduce some competing definitions of
- 4:08democracy. And then finally, I'll get into
- 4:11sort of the, the real topic AI and how it impacts democracy.
- 4:16So three points that we can make about democracy, which a lot of
- 4:23people unfortunately often confuse.
- 4:26First of all, democracy is is today limited to a percentage of
- 4:32the world. It's a fairly recent condition
- 4:35in human history. And when you say democracy,
- 4:39there's a good chance that you may not mean the same thing as I
- 4:43do when I say democracy. But then again, I mean, in our
- 4:48part of the world, a lot of us will subscribe to the old
- 4:51Winston Churchill quote that democracy is the worst form of
- 4:54government. Well, except for all the others
- 4:56that have been tried. So if I look, for instance, at
- 5:00this survey from the Economist Intelligence Unit, they say that
- 5:05less than 7% of the world population live in a full
- 5:09democracy beyond full democracies like the Nordics
- 5:14are, like Germany. Well, you know, there are flawed
- 5:17democracies like Italy, India and the United States.
- 5:22But even if we cap them as well, it's less than half of the human
- 5:27population that live in what can be described as either a full or
- 5:30a flawed democracy. What makes a democracy flawed,
- 5:33you might ask? Well, that's sort of something
- 5:36that we'll be getting into because it depends again on this
- 5:40understanding that we have of what democracy is.
- 5:46The second point that I'd like to start off with is that
- 5:49democracy is in human history, a fairly recent event, the concept
- 5:56of sort of universal suffrage. As you know, everybody over 18
- 6:00gets to vote. That is even in a country like
- 6:04Denmark, only 5 decades old. And if we go 100 years back, and
- 6:09I'm using the example of Denmark here because the timeline is
- 6:13slightly different depending on what country that we're looking
- 6:17at. But even if we go 100 years back
- 6:18in Denmark, well, we had suffrage as in the right to
- 6:21vote, but it was restricted to only 1015% of the population.
- 6:28All women were excluded and 8070% of the men of, of Denmark
- 6:33back then did not have the right to vote.
- 6:37You'll often find sort of history books saying there was a
- 6:40sort of democracy in, in Athens or the Roman Republic.
- 6:44But let's remember that in, in ancient times, you know, voting
- 6:49that was typically sort of, you know, one king, one vote, The
- 6:54Roman Emperor of the Holy Roman Empire could, could claim he was
- 6:58elected by the elected count. But, you know, that's a very,
- 7:01very limited version. If we look back sort of beyond
- 7:06the generation of our parents, democracy as we know today
- 7:11didn't really exist. And that leads me sort of to my
- 7:17main point here when we're talking about democracy is that
- 7:21even today, it's it's heavily debated what democracy is.
- 7:27And I could say, you know, that that here I'm coming up with
- 7:29with four competing understandings.
- 7:32And you have to remember that these are not sort of
- 7:38understandings of democracy that complete one another.
- 7:41They compete. And the four understandings that
- 7:46I'd, I'd like to go through, and there are more than these 4, but
- 7:49I think these four can make the point that is that we can either
- 7:52see democracy as elections, or we can see it as individual
- 7:57rights, or we can see it as an expression of equality,
- 8:02neglitarianism, or we can see it as a process of deliberation and
- 8:06dialogue. But these are not the same
- 8:08things. Why?
- 8:10How, you ask? Well, let's get into it.
- 8:12So the first way of understanding democracy is what
- 8:16some people would call the classic version of democracy.
- 8:19And that's where we simply see democracy as elections.
- 8:24So in this view of democracy, well, democracy is the
- 8:29competitive market for political power.
- 8:33Politicians, they are the entrepreneurs, voters, the
- 8:35currency. So people may not get to rule,
- 8:38but they get to hire and fire the rulers.
- 8:43Democracy works fine in this sense because it allows leaders
- 8:48to be changed without need for violence and it gives fixed
- 8:52mandates. And it works fine because in
- 8:56this view of democracy, well, we don't have to agree on
- 9:00something. And the proponents of this view
- 9:02of democracy say, well, you know, we we can never agree on
- 9:05everything. So, you know, let's have a vote
- 9:06about it. And, you know, if, if the
- 9:09majority made the wrong decision away wrong decision, well,
- 9:12there's another election coming up at some point.
- 9:16So then we have another view of democracy.
- 9:20And this is where we say, well, democracy is actually the
- 9:23protection of the minorities against the will of the
- 9:26majority. And the smallest minority is the
- 9:31individual. And, and this is the thing
- 9:33where, you know, in this at this panel here, we might, we might
- 9:37look at Core, one of the speakers coming up at the end, a
- 9:39successful entrepreneur, and say, well, you know, we should
- 9:42vote about whether Core should share his wealth with the rest
- 9:45of us. But in this view of democracy,
- 9:48Core has the right to say no. And in this view of democracy,
- 9:53democracy is actually not so much what the majority gets to
- 9:57the side, but when the minority gets to say no.
- 10:04And that is sometimes if you hear the expression democratic
- 10:07rights, well, that is also what is meant by that.
- 10:12Then there's a third view of democracy where we look at
- 10:14democracy as equality and egalitarianism, where we say,
- 10:18OK, you know, we can have votes, But if these votes are made by
- 10:22parties that are inherently unequal, as in some are very
- 10:27wealthy, some are very poor, you know, can we then say that they
- 10:32are voting about the same thing? A vote on paper is not the same
- 10:37as a vote in practice. And, and this definition, by the
- 10:40way, you know, coming back and linking it is, is one of the
- 10:43reasons that the economists, they would say that a country
- 10:48like India or the US is a flawed democracy simply because these
- 10:54are countries where people live under unequal circumstances.
- 10:59But there's a final view of democracy.
- 11:01That's that I'd like to mention also before we get into sort of
- 11:05comparing to match these these four, and that's the the full
- 11:09view of democracy where we view democracy as the liberation and
- 11:13dialogue. And this is a view of democracy
- 11:17that challenges the view of democracy as simply we vote and
- 11:21then, you know, once we voted, we can go home.
- 11:23This is the view of democracy says democracy is a fact that
- 11:26dialogue that we as citizens have with one another.
- 11:30It's a society is democratic when it forces the citizens to
- 11:37engage in dialogue with one another in order to present
- 11:42their arguments here, the concerns of the other parties
- 11:46and thereby gain a shared and common understanding.
- 11:51And that's a fourth view of democracy that clashes.
- 11:55And I think you know, as I, as I started out by by by saying,
- 11:59it's very important to remember these, these are are not sort of
- 12:03different aspects of the same picture.
- 12:08These are not sort of four parts of a overriding Uber definition
- 12:15of democracy. These are four very different
- 12:18ways of viewing democracy and indeed what we mean by
- 12:23democracy. And you could probably guess
- 12:25sort of on the right, left political divide, where
- 12:30different political parties would tend to favour, well, what
- 12:33different political parties would tend to favour what view
- 12:36of democracy, Because that's also something that we are
- 12:38constantly debating in a democracy.
- 12:44But you know, we can also say that these 4 definitions, they
- 12:49can overlap. They are.
- 12:51It's always frightfully simple to, you know, say that we can
- 12:55reduce the world into four boxes.
- 12:57Of course we can't. There's many Shades of Grey
- 13:00between them. At least 50 are here.
- 13:03But these 4 definitions of of democracy can also overlap.
- 13:07They can also help each other. But we should remember they are
- 13:09fundamentally different ideas. So some 3 decades back there was
- 13:18a American researcher called Fukuyama and and Fukuyama, he
- 13:22wrote a book called the The End of History.
- 13:26And this book was the idea that liberal democracy was the last
- 13:33offer standing 100 years ago. There was a dialogue, there was
- 13:38an ideology called communism. There was another one called
- 13:40fascism. There was also at one point
- 13:43something called Arist well, monarchy, monarchism.
- 13:47But about 30 years ago, democracy won, not because all
- 13:54countries suddenly became democratic, but because liberal
- 13:58democracy became the ideal that everyone seemed to be moving
- 14:03towards, even even places like North Korea.
- 14:07And that sort of was the was the prevailing viewpoint that, you
- 14:12know, we were living in the golden age of of democracy.
- 14:16And, you know, if we look back at the sort of history of
- 14:20democracy, one can certainly make that argument.
- 14:23But there's also people who have been arguing the opposite, that
- 14:29in fact, sort of we live in an age of democratic decline.
- 14:33And I won't throw my hat in, in the, in the lot of, of either
- 14:37party, but simply to say that even before there was something
- 14:41called artificial intelligence, our view of democracy was being
- 14:47challenged by a lot of certain by a lot of global trends that
- 14:51were happening. So for instance, we live in an
- 14:53age of globalization. This means that a wonderful
- 14:56person like Marella, she can move from Italy to Denmark.
- 14:59And but once she does that, once she does that, she doesn't have
- 15:04a Danish citizenship. She can't vote in a Danish
- 15:06election here. So when we live in a globalised
- 15:09world where people, they change the country in which they live,
- 15:13well, the right to vote all of a sudden becomes restricted.
- 15:17And in in countries like like Italy, like Denmark, like the
- 15:22US, some 10 to 15% of the of the population that's living there
- 15:27will now not have citizenship and therefore will be restricted
- 15:32in their voting. Of course, this is where a
- 15:34country like Luxembourg goes. Hold my whistling.
- 15:39I believe there's something like 70 or 80% of the population in
- 15:42Luxembourg, which is a member of the European Union.
- 15:45They do not have a citizenship and does not have the right to
- 15:50participate in their election, which is an extreme example.
- 15:54I, I, I take it, but it's one of the ways in which globalisation
- 15:57challenges our view of, of democracy.
- 16:00In the same way, you can have many views about taxation, but
- 16:05you have to acknowledge that taxation levels across all
- 16:10countries have risen dramatically over the last 50
- 16:13years. And this leads to sort of an
- 16:16interesting situation where Western countries now need to
- 16:20tax more areas because they're taxing some areas quite heavily
- 16:25in order to avoid sub optimization.
- 16:27So that's challenging. Well, at least the individual
- 16:30rights part of it of democracy. There was a an interesting case
- 16:36from Germany some 30 years back where the German Supreme Court
- 16:42said that if we ever were to, if the German government ever was
- 16:46to tax somebody above the threshold of 50% for a given
- 16:49activity, well, that's no longer taxation, that's appropriation.
- 16:52And the German government can still do it, but they would have
- 16:54to follow a different procedure. But a decade went by and then
- 16:59the German bundles for Fassenstein, they've reversed
- 17:04the ruling simply because it didn't seem political.
- 17:08Well, I don't know why they reversed the ruling, but they
- 17:10had the arguments. But as an example of how some
- 17:13people to take the view of of democracy would say that
- 17:18democracy in decline, I present that as an argument.
- 17:21And then we have the thing about equality because we have again,
- 17:25globalization here, which leads countries to become more equal
- 17:30in terms of economics. But at the same time, we see
- 17:33rising inequality within a country.
- 17:35And this is also related to automation because of
- 17:38information allows talent to to thrive.
- 17:41Well, people who are talented will sustain higher incomes, but
- 17:45they but what if they no longer need to live next to the people
- 17:49who are, for lack of a better term, less talented?
- 17:52And finally we have the whole sort of social media wave we're
- 17:56talking about, sort of a shared reality.
- 17:59Well, social media algorithms tend to reward virtue signalling
- 18:03rather than honest deliberation. It's the got you comment that
- 18:08gets the likes and the views, not the I understand what you're
- 18:13saying and I agree to disagree but yadda yadda yadda.
- 18:17So the argument, the case can be made that even before AI entered
- 18:22the chat, the question was already where's democracy at?
- 18:27And that leads me sort of to where I finally get to the link
- 18:30up between AI and democracy. Because depending on how we view
- 18:35democracy, then AI can either help democracy or it can harm
- 18:40democracy. And likely it will be doing both
- 18:43things at the same time. So if we view AI as elections,
- 18:50well, AI can simplify complex information.
- 18:53It can make voting choices easier, easier to understand,
- 18:56but oh, who is holding the finger and the weight?
- 18:59Who is deciding what gets shown? And that is how AI can obviously
- 19:06harm democracy. If we view democracy as
- 19:10individual rights, well, you know, AI is a fantastic tool to
- 19:13help citizens understand and and claim their rights.
- 19:17But if AI is used by the powers that be, this is where we begin
- 19:22to lack the accountability that we take for granted in a
- 19:26democratic society with equality.
- 19:30You know, we can also imagine a case where AI will help
- 19:36democracies by lowering the barriers to democratic
- 19:40participation. If you lived in Denmark but
- 19:43didn't speak Danish well, you could have AI translate Danish
- 19:47election material for you. But on the other hand, AI is
- 19:52also a question of who gets to access what tools.
- 19:56And if we all of a sudden begin to see a gap between, some
- 19:59people will have access to fantastic superior models of AI
- 20:04and others won't, well, that will increase inequality.
- 20:09And finally, sort of with the liberation and dialogue.
- 20:12AI can help us, you know, present our arguments in a
- 20:15better way. But it can also be the case
- 20:17where we see that social media, the Internet will be flooded by
- 20:22synthetic noise. That seems like a lot of people
- 20:25are participating in a democratic debate.
- 20:27But in reality, it's just machines spewing out whatever,
- 20:32you know, machines do. And if we go deeper into that
- 20:36final point, then there's the big question of when we see
- 20:40something, you know, on the Internet, is it created by
- 20:45human? Is it made by an AI?
- 20:47There's an interesting sort of overview here, which is a year
- 20:50old now, which says all of the things you encounter on the
- 20:54Internet, roughly half of it will be made exclusively by an
- 21:00AI. But even the split between human
- 21:03AI generated, well, AI is challenging that isn't it?
- 21:07Because AI can be present in many different places in the
- 21:12creation process. And when, when I show these
- 21:15slides to my students and, you know, ask them to guess, you
- 21:19know, where we at today, some of them say we're, we, we must be,
- 21:22we must be past the 80 or 90% mark.
- 21:26I don't know. But this is this is where we're
- 21:29looking at it. And what we just have to really,
- 21:32really remember is there's a question here of reality versus
- 21:40what the AI will show you. Yeah, this is back to the point
- 21:44that the late Professor Daniel Kahneman, he made in his book
- 21:50Sinking Fast, Sinking Slow, where he introduces the term
- 21:54whizzy at what you see is all there is, which is the bias that
- 21:59what is shown, the answer that the AI shows us is a reflection
- 22:04of reality. And it's not a reflection, it's
- 22:10a selection. Yeah, it's a part of reality
- 22:13that was shown is is selected based on what data the AI has
- 22:16been fed with, what rankings have been provided, what the AI
- 22:21believes is the right answer based on your history, what the
- 22:24the owners of that AI have put in.
- 22:26And all of this, you know, shows us just, you know, something
- 22:30that is very selected. But The thing is that due to the
- 22:34concept of variance, which again is one of those things that in
- 22:37my experience, people, even people who use AIA lot, they
- 22:41don't really get this. But variance is is is is very
- 22:45basically that an AI will never give you the same answer twice.
- 22:50I mean, the, the Greek philosopher Heraclitus said, you
- 22:55know, a human can never cross the same river twice because the
- 22:58water will not be the same, nor will the man be the same.
- 23:01But today, I mean, if you ask an AI the same prompt, it will vary
- 23:06it's answer. And how much will it vary?
- 23:09Well, one little experiment that the nice fellows over at Spark
- 23:13Tura make was that they found that you needed to ask Claude
- 23:18the same prompt 1428 times before it gave 2 answers with
- 23:26the same list of brands. You know that you were that you
- 23:30were looking for. Now this this variance also
- 23:35strangely enough, seems to be unequally divided.
- 23:42For instance, when you ask the AI about books, it, it presents
- 23:45a much more either variety of different answers than if you
- 23:51ask it for say a Volvo dealer. But but then again, and this is
- 23:55where variance and the selection can trick us, right?
- 23:58Because if we're seeing different answers all the time,
- 24:00we can say, OK, well, now I know the 9 Volvo dealers that there
- 24:05are, I can just prompt more. Yeah.
- 24:07But we should remember that, that in this case, sort of, you
- 24:09know, if people in the US, they were, they were doing a test
- 24:14and, you know, asking, asking ChatGPT and, and the other AIS
- 24:19about, you know, give me a, a Volvo dealer in the greater Los
- 24:22Angeles area. And they got the same 9
- 24:25dealerships. But there are roughly 300 Volvo
- 24:29dealerships, you know, in, in the vicinity.
- 24:35If we really take sort of the greater California area, only
- 24:38those 9 here could mention why only those nine?
- 24:41What happened to the others? And this is a great example, I
- 24:44think of the, the way that AI takes out a, a slice of reality
- 24:53and shows it to us. I mean, I could have made the
- 24:56same example, you know, based on unique sci-fi novels.
- 24:59But the mind there would be mind boggling because even if the
- 25:02variance there is, is higher and the answers, well, the, the
- 25:05potential market, both of the potential things that could be
- 25:07mentioned, that would be even larger, wouldn't it?
- 25:10And this comes also back to sort of when we're talking about AI
- 25:14and democracy. There was a, a study made by the
- 25:18Danish Centre for AI Research by Stephanie Brundle, where she
- 25:24came up and asked sort of there was a election in Denmark here a
- 25:28couple of months ago. And, and she came in and, and
- 25:33asked Chuck DPT, who would you recommend?
- 25:35What party would you recommend to vote for?
- 25:37And then it would never give the same answer twice.
- 25:40But consistently the Danish Unity Party was at the top of
- 25:45the ranks. And for those of you who don't
- 25:47know much about Danish politics, that's the left most party that
- 25:52there is. So why is is the great question?
- 25:59I mean, were the Danish communists the best at
- 26:02optimising, you know, their web page in order to be found?
- 26:06Is it a question of that the, you know, they, they, they may
- 26:09want to refer a capitalist society, but they certainly have
- 26:12the search engine optimization tools to, to do it or, or what
- 26:17is happening here? Is it the reinforced learning of
- 26:22ChatGPT that's giving it this slant?
- 26:26And is that the reason, for instance, that Gemini had a
- 26:29different ranking? But you should also note, I
- 26:31mean, there's some parties here, they're never ranked first.
- 26:33The Danish Conservative Party, which got a fair election
- 26:36result, They were never mentioned as the first choice by
- 26:40by either of the two models that were tested here.
- 26:43And and this is a single great sort of example of what are some
- 26:46of the potential problems that we are looking into.
- 26:49Of course, you might say the potential problem here is if
- 26:52anybody ever asks this question, you know, what should, what
- 26:57should I vote for? You know?
- 26:59But then again, even if people don't ask it, AI will provide
- 27:04other tidbits of information that will help shape our sinking
- 27:08before we even get to that. And I think we're looking a
- 27:12little bit into sort of there's an emerging science here of, of
- 27:16how we write so that we are found by the AI.
- 27:22So if if my theory here that the Daniel Unity party there was
- 27:26simply very good at writing in a way that the AI liked and that's
- 27:29the reason they're the top of the rankings, well then it's
- 27:32because they did something like this.
- 27:33They wrote with clear summarisation.
- 27:35They wrote with AQ and a format. They wrote with structured
- 27:39sections. Maybe, maybe, maybe.
- 27:44And also I'd, I'd like to sort of also just bring this slide
- 27:48here into the, into the mix because I know at least one of
- 27:51the people speaking later today is, is very fond of Claude.
- 27:55Claude is the rich man's AI out of US households with an income
- 28:00over $100,000 a year. Well, 80% of of those people,
- 28:08they use Claude, whereas Meet AAI is the poor man's AAI
- 28:13apparently. Why?
- 28:15I mean, because if you go in and look at these different AAI
- 28:18models, their pricing is not all that different.
- 28:21So what caused this slant? And that's sort of, I think one
- 28:26of the questions that we should explore later today.
- 28:30So anyways, that's what I had here for you and Mirela.
- 28:37I hope that this presentation here with the different views of
- 28:41democracy and how AI can help unhinder based on those 4
- 28:47definitions or understanding of understandings of democracy will
- 28:50be a good start of point for the other people that you invited
- 28:55today. Sorry if I have stumbled earlier
- 28:58in my introduction in my intro because they're apparently it
- 29:02was a technical issue streaming on Facebook and unfortunately
- 29:07sometimes might happen those kind of stuff, but people can
- 29:11still attend to the live event on LinkedIn and on YouTube.
- 29:17I'm sorry, but again, things, this kind of stuff sometimes
- 29:21happens. So yeah, this is what I wanted
- 29:25to to say. Mark the floor, it's yours.
- 29:30Thank you. Thank you.
- 29:33Difficulty being follow on such a such great presentation.
- 29:41I think we're used to give you an introduction to what
- 29:46democracy is. I'd like to dive a bit more into
- 29:50some of the things that he also talked about, about what AI can
- 29:55do, what kind of risks we're facing with democracy.
- 30:01And I'd like to start with a small observation that I think
- 30:04most of us recognize, even if we've never really put a name to
- 30:09it. We've been having a lot of
- 30:11conversations about what AI can do, how fast it is, how capable
- 30:16it's becoming, what jobs it might change, and all of that
- 30:21matters. But a question I keep coming
- 30:23back to is a different one. Not what AI can do, but what it
- 30:28quietly does to the way we govern and who we can hold
- 30:32responsible when it goes wrong. Let me start with something
- 30:40basic, something we often take for granted.
- 30:45Democracy is not just a voting mechanism.
- 30:48When we strip it down to its foundations, it rests on three
- 30:52things. Votes, yes, but also
- 30:55explanation. The ability to know why a
- 30:59decision was made and to be able to challenge it, and
- 31:04responsibility somewhere in the system.
- 31:07A person, not a machine, who can be held accountable when
- 31:12something goes wrong. These three aren't bureaucratic
- 31:17niceties. They are the conditions under
- 31:20which citizens can trust that the institutions acting on their
- 31:25behalf are actually answerable to them.
- 31:30And here's the real risk I want to name today.
- 31:33Not that AI takes over governments, but that
- 31:36institutions quietly surrender accountability because it became
- 31:41convenient to. Here's what I keep coming back
- 31:50to in my work at a potatoon. AI isn't just changing how
- 31:55governments work, it is becoming the infrastructure through which
- 32:00the state governs. The mechanism of governance
- 32:04itself is shifting. Look at where AI has already
- 32:09arrived and note that these are not peripheral functions.
- 32:14These are the points where the state most directly touches
- 32:18citizens lives. Welfare benefit assessments,
- 32:23Immigration and visa decisions. Predictive policing and risk
- 32:28scoring. Tax authority case processing.
- 32:33Child welfare prioritisation. Municipal service allocation.
- 32:41In many of these cases, AI systems are now either making or
- 32:46heavily shaping consequential decisions without public debate,
- 32:51and often without the institution itself fully
- 32:54understanding why. When decisions are made by
- 32:58systems that are both complex and partially opaque, the
- 33:03question is not only what was decided, it is who decided and
- 33:10whether anyone can be held responsible.
- 33:19Now, if you've seen the other instalments of of Of AI and the
- 33:25Human Future, you will recognise this framework that I set up.
- 33:31But it bears mentioning again. And for those of you who haven't
- 33:36seen the other instalments, let me just read restate this
- 33:43because I think this makes what I've just been saying more
- 33:46tangible. I used to talk at one of the
- 33:54previous talks. I talked about AI and leadership
- 33:57judgement. The same logic applies, yeah,
- 34:00but scaled up to the institutional level.
- 34:05At the first level, as you can see AI as an administrative
- 34:10assistant, this is the safe zone.
- 34:12This is where it summarises cases, searching precedents,
- 34:16structuring documents, drafting correspondence.
- 34:21This is the zone where you can use AI freely.
- 34:23This is problem free, so to speak.
- 34:26Almost. This is where it generally saves
- 34:29times without touching democratic accountability.
- 34:34At the second level, AI can become a policy advisor.
- 34:39It's the more active song risk assessment, resource allocation,
- 34:45outcome forecasting, policy modelling, powerful and
- 34:50genuinely useful. But it requires a human in the
- 34:53loop on every consequential recommendation.
- 34:57The official must still own the judgement of a third level and
- 35:04this is where things can become problematic.
- 35:07AI can become the decision maker.
- 35:11I call this the danger zone. This is areas like welfare
- 35:16eligibility, visa outcomes, risk scoring, injustice and child
- 35:21welfare. When AI concludes alone and no
- 35:25one in the chain can fully explain the decision,
- 35:29accountability disappears. And this is where democracy
- 35:33fractures. And where am I going with all
- 35:39this? The point here is I think that
- 35:42we are seriously in danger of having drifted into what I call
- 35:46Level 3. And this is problematic when
- 35:53it's not based on democratic decision.
- 35:58And the phrase drifted here is used deliberately.
- 36:02This isn't a conspiracy I'm talking about.
- 36:04Nobody decided that AI should replace democratic
- 36:08accountability. It happened through thousands of
- 36:12small procurement decisions, convenience choices, and
- 36:15unchecked assumptions. And this is actually the harder
- 36:21problem. It's easier to resist something
- 36:24that announces itself. Much harder to resist the quiet,
- 36:28gradual shift that always seemed reasonable one step at a time.
- 36:34Now you can argue that this is this is not necessarily the
- 36:38case, or at least not the case in in every case.
- 36:42But it's a serious risk and we are getting there in my opinion.
- 36:52So let me make this more tangible.
- 36:56I'd like to walk you through a scenario that we have seen
- 37:02variants of in real life. A citizen receives a rejection
- 37:08on a welfare application. They don't understand the stated
- 37:12reason. They call the case officer.
- 37:16The case officer doesn't know exactly why the system flagged
- 37:19it. They escalated to the
- 37:21supervisor. The supervisor lacks access to
- 37:25the model decision logic. They recommend an appeal.
- 37:30The appeal board cannot order the algorithms waiting.
- 37:34They can only review the inputs, not the process.
- 37:38The ministry when asked, points out that the Commission, they
- 37:42commissioned the system but they do not own the model.
- 37:46The vendors code is proprietary. Now follow that chain.
- 37:52No one lied. No one was negligent, no one
- 37:56failed to do their job. And yet no one, not a single
- 38:01person in the entire accountability structure, can
- 38:04fully explain why this decision was made.
- 38:08This is not a technical failure. This is a democratic failure.
- 38:20And here is the paradox at the heart of all of this.
- 38:24We're asking citizens to trust institutions whose decisions are
- 38:28shaped by systems those institutions cannot fully
- 38:32explain. Audit or replace.
- 38:36That is not an institutional trust.
- 38:38That is asking citizens for blind faith.
- 38:41And there is a crucial difference.
- 38:44Institutional trust is earned through accountability.
- 38:48It is built on the understanding that if something goes wrong,
- 38:51there is a process, a real 1, not a symbolic one, by which
- 38:56citizens can know what happened, challenge it, and expect a human
- 39:00to stand behind the answer. This matters in three concrete
- 39:05ways, which you can see on the slide.
- 39:09When institutions cannot explain their decisions, the discipline
- 39:13see erodes, regardless of whether outcomes are actually
- 39:17correct. People don't trust what they
- 39:20can't understand. When opacity becomes structural,
- 39:26the right of appeal is destroyed.
- 39:28In practice, even when it technically exists in law, you
- 39:33cannot challenge what you cannot understand.
- 39:36And when the state governs through systems it doesn't own
- 39:39or fully control, sovereignty itself is being quietly
- 39:43transferred not to a foreign government, but to a vendor.
- 39:54Now, I don't want to leave you with a picture of AI governance
- 39:58as pure crisis. That would be incomplete.
- 40:02And it would miss the most important point.
- 40:04The question of which values AI systems should follow is not yet
- 40:09answered. It is open, but it is being
- 40:12answered right now in server rooms and procurement rounds,
- 40:16not in parliaments and citizens assemblies.
- 40:19The good news is that there are real examples of it being done
- 40:23differently. Taiwan's The Taiwan process uses
- 40:28AI as democratic dialogue infrastructure, not to replace
- 40:33the liberation, but to surface genuine agreement and expose
- 40:37real disagreement. The AI serves the democratic
- 40:41process. The democratic process is not
- 40:44replaced by the AI. It is transparent.
- 40:48It is publicly audible. Citizens shape the output.
- 40:54Another example of this is Barcelona's Decidum platform.
- 40:58It shows that democratic ownership of digital
- 41:02infrastructure is possible. Open source code, citizens
- 41:06proposing changes to the platform itself.
- 41:09The city, not the vendor, owns the data.
- 41:13These are not utopian gestures, they are operational and they
- 41:17have been running for years. The AI alignment question is not
- 41:22technical, it is democratic, and the window to shape it is open
- 41:28for now. Let me close my segment with a
- 41:37practical tool, something you can actually use.
- 41:41Whether you're a citizen, civil servant, journalist, or someone
- 41:45who procures technology for a public institution, everyone in
- 41:50the technology world talks about their stack, the infrastructure
- 41:54their systems run on. I want to introduce a different
- 41:58kind of stack, the accountability stack.
- 42:02Take any AI assisted decision in your organization or in your
- 42:07public life, A welfare assessment, a visa decision,
- 42:11risk score, case prioritization and walk the chain.
- 42:16Layer one. The person affected, can they
- 42:20understand the stated reason? Layer 2.
- 42:25The caseworker or officer, can they explain the full basis of
- 42:29the decision? Layer 3, the supervisor or
- 42:34manager, can they access the decision logic?
- 42:39Those three are the top of the stack, the layers where
- 42:43accountability should be intact. Then you hit the gap.
- 42:48The appeals body. Can they audit the models
- 42:51waiting? The institution or ministry?
- 42:55Do they own and control the system they govern through?
- 43:00In most organizations using AI at Level 3 today, the stack
- 43:05breaks somewhere around layer three.
- 43:07That is your problem. That is where the democratic gap
- 43:10lives. Where the stack breaks.
- 43:13That is where your democratic problem lives.
- 43:17The exercise is simple. Pick one AI assisted decision in
- 43:21your world, Map the stack, find where it breaks.
- 43:26That is your starting point. I'll leave you with the
- 43:35statement I keep coming back to, the one that I think actually
- 43:39captures what is at stake. In a democracy, accountability
- 43:44is not optional. If no one can explain the
- 43:47decision, no one can be held responsible for it, and this is
- 43:53a democratic failure. Thank you.
- 43:59Thank you so much, Mark. Thank you.
- 44:02Let's go to the next guest, which is Hello Welcome.
- 44:15Hello, Mirela, and thank you. You are welcome.
- 44:18I'm sorry but it seems like today is not my day.
- 44:21I keep, you know, you're doing mistakes and mistakes.
- 44:25We're, we're live and and and and the, the IT, it's also very
- 44:30interesting. So I mean, technical ditches are
- 44:33just technical ditches. Yeah, definitely.
- 44:38I cannot see your presentation though.
- 44:41No, but if I share my screen, I think we'll we'll get there as
- 44:46well. So let me do just that and this
- 44:55way round I will share the actual presentation and I.
- 45:00Guess is this. Well, I'm I'm it's coming up.
- 45:07Is my screen visible now? Yes.
- 45:11Perfect. OK, so I'm going to let you to
- 45:14your presentation. The floor is yours.
- 45:16Thank you, Karl. Thank you so much and and thank
- 45:20you all for for for digging into this topic.
- 45:24It's so interesting and that's really so much to, to, to do.
- 45:28And my pleasure today is really taking it the next step because
- 45:35as Yust was mentioning, well, we have a starting point, as Mark
- 45:43was telling, we dig deeper in part through the technical
- 45:49parts. And my presentation will very
- 45:52much be about the, the, the, the technical parts.
- 45:56So for myself, I enjoy so much being every citizen.
- 46:04I depend on others in being that.
- 46:08And to a larger and large extent I depend on technology.
- 46:13And that technology is really often made as an infrastructure.
- 46:19It's a lot of technology connected, the Internet for
- 46:23start, services for continuations and AI built on
- 46:28top of that. And if that stack is not made in
- 46:36your knee area, if you don't control that infrastructure, it
- 46:42might be a topic, might even be a problem.
- 46:46And in the case of democratic institutions, it seems that in
- 46:52many countries for star Denmark where I'm located are really
- 46:56outsourcing the digital infrastructure to other nations
- 47:01that could have all the interests and all the agendas
- 47:06need not be a problem. But it is definitely a a theme
- 47:12and we are for sure in Denmark, probably in other countries
- 47:19really living our life on least ground in the Danish market.
- 47:26That will often be American technology.
- 47:29I know there is technology from other countries out there and I
- 47:32know that other countries make different decisions for the good
- 47:38and for the well total package. And I think we should start out
- 47:45from what is the interest of the national state, what is the
- 47:50interest on a national scale? Because there might be
- 47:54interests, you might get your messages from basically another
- 48:00country. And as long as that country is a
- 48:04close ally, it's one situation. If things changed in the world
- 48:08situation, and at times it does well, it might become a bigger
- 48:14topic. And as so many services are
- 48:17interconnected, it might not be an easy task to switch to a
- 48:20different ecosystem. And we are moving down such a
- 48:27path because by now and these in Denmark where I'm located, we
- 48:34run to a very large extent on Microsoft technology.
- 48:37That goes for the Windows operating system, that goes for
- 48:42search engines, that could also be Google and I'll get back to
- 48:45that. That go for for the Office
- 48:49software, Word, Excel and so on. And it need not be a problem
- 48:53because that's really just supporting technology.
- 48:56But when we build on top AI, in the case of Microsoft, that
- 49:00would be cope pilot. Well, it has a certain
- 49:04direction, it has a certain functionality and it is an
- 49:07American service, need not be a problem, but it is definitely a
- 49:11theme and there are other services out there.
- 49:14There are European ones, but if we still access it from Windows
- 49:19computer or a Mac computer, on both cases there are American or
- 49:23from an iPhone or from an Android device, in both cases,
- 49:27American software. Well, we are not totally
- 49:31European and indeed not be a problem, but it is definitely a
- 49:35topic and the terms of service are often not even negotiated
- 49:43because, I mean, it might be a possibility that you can
- 49:47negotiate that the data sender should be in a certain location.
- 49:51Sometimes you will succeed in that, but then the physical data
- 49:55sender is still connected to the rest of the world, which is a
- 50:00topic and and the legal terms can be defined.
- 50:06They can be changed afterwards and you might notice if they're
- 50:10changed. You might have a reasons to
- 50:13object about it and whether or not you are at the end of the
- 50:18day served. At least in the case of Denmark
- 50:21being a small country in a big world, well, that remains to be
- 50:25seen and all of this need not be a problem, but it is definitely
- 50:29a topic and we might end up being really the cloud serve
- 50:38Mark was getting a little down that path.
- 50:42And I mean, if we labor on land that is known, not owned by us
- 50:49in a digital perspective and if we are generating value that
- 50:55goes to big tech companies located in other countries And
- 51:00if we are really bound to those systems because it's so complex,
- 51:04if you want to break out of those systems, well, are we
- 51:09still free people? Are we still serving our
- 51:13democracy in a good way? I hope so.
- 51:16But sometimes I have my thoughts, whether we are moving
- 51:19in a direction that is not ideal, so to speak.
- 51:25But well, let's see. And it's not only about
- 51:33Microsoft, it's not only about Copilot.
- 51:36It's also about the search engines.
- 51:39In Denmark and in many countries, Google is the biggest
- 51:43search engine. Google is presenting results.
- 51:48Up till last year 2025, it was usually a number of ads and then
- 51:53a number of results and you could pick and choose and find
- 51:56the best one, the most trustworthy ones and you could
- 51:59probably see who was the voice behind it and what could be
- 52:03their interest. Now we are little by little
- 52:06seeing an AI generated answer and it might not be clear what
- 52:13was the source, it might not be clear what is the interest.
- 52:17And it is an American service. Many users are not going further
- 52:22down to check the background sources. 60% and probably even
- 52:27more according to the the research that I've done, many
- 52:31will really see the one truth presented by these machines.
- 52:36And in some cases consequences are not that big.
- 52:38I mean, if I am to buy a new, well, I don't know if cheap
- 52:44electric device, well, I will buy some cheap electric device
- 52:48and I'll be fine. But the overall topic for, for,
- 52:51for this session is about democracy.
- 52:53And sometimes it does affect our democracy and again, need not be
- 52:59a problem, but it is your topic because these services are
- 53:03interconnected. These services are to a large
- 53:07extent really black boxes. And they might be changed by
- 53:11actors in that market and you might not know what is their
- 53:16exact reason for making that change.
- 53:19And it seems to just scale and scale and getting bigger because
- 53:25those tech companies are so big, they're so influential.
- 53:29They are so interconnected with interests that might not be
- 53:34democratic, that might not be located to the actual country.
- 53:38So I guess it is a topic and it is also a topic for the
- 53:44individual citizen because you really need to train your
- 53:49cognitive muscle. And I suggest also going
- 53:55offline. I suggest also reading books.
- 53:57I suggest also talking to actual humans.
- 53:59This is not only digital world, but it is to a large extent a
- 54:03digital world. And that digital world is
- 54:06connected to the democracy both from the citizen perspective and
- 54:11from the authorities perspective and really connected to
- 54:15technology, that is, that is not a local.
- 54:20And I would not call this corruption.
- 54:26I might use the term soft corruption because in the case
- 54:32for me as a Dane, being a citizen in a small country,
- 54:37being dependent on services and other countries, which need not
- 54:42be a bad thing. Well, it might be that little by
- 54:46little we get an inference, we might get influenced by all the
- 54:52values and from biases, and it might not even be through bad
- 54:57intentions, but it is part of the package.
- 55:02And this goes for other countries as well.
- 55:04I mean, this session is not only about Denmark.
- 55:06This is the where I'm located. I can see that in Germany they
- 55:10do have their own tech companies and sometimes they make
- 55:13different choices than than American software, but it's hard
- 55:17to replace the entire system. The same goes for other big
- 55:22European countries. It goes for Italy, it goes for
- 55:25France. I'm not into every detail, but I
- 55:28can see that there are players out there and it might be that
- 55:31they become so strong that Denmark might even adapt there.
- 55:34That technology. It hasn't happened at least to a
- 55:38large extent so far, but who knows what the the future will
- 55:43bring. And it is a topic because it is
- 55:46about the training data, it is about optimization target, it is
- 55:50about how it is moderated. It is also about influence from
- 55:56countries that are located far from Denmark.
- 55:58And it is well described that the US has it has its interests.
- 56:05It is clearly described that China has its interests and once
- 56:09again need not be a problem. But it is surely a topic, and in
- 56:18a traditional democracy, there are the democratic institutions,
- 56:25there are also the media. We would often say that the
- 56:29media could really be the 4th power or the 4th estate or
- 56:33however we put it. In Denmark, it's Denmark's
- 56:37radio, the abbreviation would be Dr. In Denmark, it would be R&D,
- 56:43in Italy it would be high. And then other countries, it can
- 56:47be other institutions and many European countries have strong
- 56:51institutions. Many Europeans countries have
- 56:55strong media that should be at least a independent, that should
- 57:00be true to the value of that country, that should support the
- 57:06the democracy in that country. And once again, the democracy is
- 57:10our Oval topic. But if we don't get it directly
- 57:16from that media, if we get it through an AI service on large
- 57:21language model, well, we do access that content through an
- 57:26American service or from another country.
- 57:30And it might tell it word by word how it was originally
- 57:35written. And it will more likely not tell
- 57:39it word by word. It might short it down.
- 57:42It might have have a different angle and again, it is a topic
- 57:47not necessarily a problem. And if we feed in that
- 57:55information well, we're better off if in then if we don't feed
- 57:59in that information. But then again, big American
- 58:03services, a small country and public service institutions that
- 58:09are big at least in the local market, but maybe not on on an
- 58:14international scale. And we might end up in a decay
- 58:18loop really. And it is usually not happening
- 58:23overnight. It is really one day at a time,
- 58:261 system update at a time, one prompt at a time, 1 new
- 58:31generation at a time. And in the short horizon, at
- 58:36least from where I see it and from where I'm located, we do
- 58:40have a strong democracy. We do have strong institutions.
- 58:43We do have gatekeepers. We do have services that don't
- 58:48depend on foreign technology. But it seems to move in that
- 58:52direction where day by day we depend a little more on
- 58:56technology and a little more on international services that
- 59:00might have a different starting point, that might really have a
- 59:04different area of of interest. So I think that we should really
- 59:11care about these, these media or these 4th state, because the
- 59:18risk is that we abolish it. The risk is that we really give
- 59:23away what has been built up, what is has been grown strong,
- 59:28which has played a role in the local society.
- 59:31And once it's no longer there, well, will we ever get it back?
- 59:34I don't know. And there are European
- 59:41alternatives out there. There are the, the, there is
- 59:46the, the, the large AI company, the French one called Mistral.
- 59:51And, and it is a possibility. I know that it's strong in the
- 59:55local French market. I know that there are the
- 1:00:00systems in Denmark where I'm located, there is the always 2
- 1:00:05system. And it is possible to go down
- 1:00:10that path whether or not we choose it.
- 1:00:12I guess that's a decision. And I can surely see that it's
- 1:00:17tempting to just stay in an ecosystem that already contains
- 1:00:22an operating system and an office package and intelligence
- 1:00:25build on top that can collect data from the mailbox and the
- 1:00:29calendar and everything. I use it myself, and I have a
- 1:00:32big level of convenience in it. I don't see big problems to it,
- 1:00:36at least on the short horizon. But it is a topic and and it is
- 1:00:42a topic that could really affect whether we will stay sovereign
- 1:00:49as a nation and as a citizen. The world is connected for the
- 1:00:53good and for, well, the total package.
- 1:00:56And it might be that, that someday the Danish society or
- 1:01:00whatever country should go down a path like the French one with
- 1:01:05local technology, like the decode project in Barcelona,
- 1:01:10like in the, the, the, the Taiwan approach, which is not
- 1:01:13European. But it's still worth taking,
- 1:01:17taking inspiration from. I'm not the one to give the
- 1:01:22final advice, but I hope I'm the one to at least give an overview
- 1:01:27of what choices we can make and what could be potential
- 1:01:32consequences of that. And I know that after me there
- 1:01:35is another speak after me, there is a question and answer
- 1:01:38section. And I hope that will be our
- 1:01:40opportunity to take, even give up.
- 1:01:43But while I'm still presenting, I have some advice.
- 1:01:48Because they shouldn't get dark. There are definitely topics,
- 1:01:51There are definitely decisions to make, but there are also
- 1:01:56advice. At least this is my best chart.
- 1:02:01As long as you use AI with intent, well, you're better off
- 1:02:04than if you're not aware if you triangulate across models.
- 1:02:10I often do that from me. I do use technology from
- 1:02:14Microsoft. I also use technology from Apple
- 1:02:16and from Google and from other players, mainly American ones.
- 1:02:22I, I, I have to admit that, but I sometimes use Danish
- 1:02:27technologies, I sometimes use European technologies.
- 1:02:31And I would so much like to see technologies from local vendors
- 1:02:37being becoming so good and so well integrated that it is a
- 1:02:42real alternative from where I see it, at least from a
- 1:02:46convenience perspective, from a market perspective.
- 1:02:49I mean, I, I, I do earn my money as an IT consultant.
- 1:02:52So it's also about customer demand.
- 1:02:56I live quite well from Microsoft technology, but I could live
- 1:03:00just as well from other technology if demand is there.
- 1:03:03And I hope that demand will get there.
- 1:03:06And it is also about doing the audit.
- 1:03:08Yes, you can use technology, Yes, you can have the
- 1:03:11convenience and yes, you can have the productivity.
- 1:03:14And yes, you can have so many advantages.
- 1:03:17And I hope you can do it and still say staying true to being
- 1:03:21the free citizen in the well functioning democracy.
- 1:03:25But because then once again, that's really what is the
- 1:03:29overall theme of, of of this wonderful YouTube live.
- 1:03:34And this is where my presentation is little by little
- 1:03:40wrapping up. I'm the third out of four and I
- 1:03:45enjoyed so much what was before me.
- 1:03:48I look so much forward to what is next because Charleston has
- 1:03:53so much to say about the nervous system.
- 1:03:55So Charleston has so much to say about the human perspective to
- 1:03:59this. And I think that when handing it
- 1:04:02over, I think that we can get from my fairly technical
- 1:04:07perspective connected to democracy to a very human
- 1:04:11perspective connected to democracy.
- 1:04:14So thank you for for giving me the this opportunity.
- 1:04:19And I think I will stop my screen share and hand it back to
- 1:04:26Yumi Rella for start and then on to Austin the next.
- 1:04:30Yeah, thank you. So our next guest is Chosen.
- 1:04:35Let's see if he's ready. Hello Chosen.
- 1:04:39Hello, Marella, and thank you for having me here again.
- 1:04:43And so I tried to bring out my presentations and while I'm
- 1:04:51doing this. Let's see.
- 1:05:01I'm here to help you. I think I managed today.
- 1:05:06It seems like it's not my quite my day.
- 1:05:08I don't believe in this kind of stuff, but literally it's like
- 1:05:11I'm keeping stunning things and technology is not quite yeah my
- 1:05:19#1 fan. Unfortunately today I don't
- 1:05:22know. Might be the vibe, who knows.
- 1:05:25Who knows? Who knows?
- 1:05:27Can you see anything? Yes, I can see your
- 1:05:31presentation. Here we go.
- 1:05:34I'm going to present them chosen, so the floor is yours.
- 1:05:38Thank you. Yes.
- 1:05:39Thank you, Thank you. You're welcome.
- 1:05:42And to start with I must see say thank you for call for the pre
- 1:05:46presentations. But I will keep this simple
- 1:05:50because the subject is so much bigger than a few minutes.
- 1:05:56So here we go when when we speak about AI and democracy, we often
- 1:06:03focus on technology regulation and misinformation.
- 1:06:09And I want to focus on the psychology layer, trust,
- 1:06:16judgement and the human ability to stay on orientated because if
- 1:06:27that layer weakens, democracy weakens from the from the
- 1:06:33inside. So if democracy was a nervous
- 1:06:40system, what is happening to to it right now?
- 1:06:50Well, my layers are you my layer as I say the way I will not
- 1:06:57speak about the the law, the platforms or the governance by
- 1:07:02it all, rather want to speak about the psychology layer
- 1:07:11beneath them under this. So democracy depends on
- 1:07:19attention, on the trusts and emotional regulation, perception
- 1:07:28and of course judgement. You know, it's very, very
- 1:07:33important. But the question I have for for
- 1:07:39this one is what happens to democracy when the when the,
- 1:07:48when the when the human layer becomes unstable?
- 1:07:57Next month, Well, democracy is only it's not only a political
- 1:08:08system. No, it's not only a political
- 1:08:13system outside outside us, you know, it's also requires in a
- 1:08:20capacity inside the the person to stay oriented to tolerate,
- 1:08:33you know, the uncertainty without the doubt or collapsing
- 1:08:42to trust the what do you call it navity.
- 1:08:48Navity, you know nativitate and Danish and to respond rather
- 1:08:59than you know better to respond rather than than to react.
- 1:09:06Respond, not react with ourselves.
- 1:09:12Capacities, democratic judgement.
- 1:09:17Well, it becomes fragrant. And, you know, this is very
- 1:09:24important to understand. So can democracy function if
- 1:09:30people lose any capacity to think before, before they react,
- 1:09:37You know, they're very fast. Yeah.
- 1:09:44It's going to be very, very yes. So the use was go into it and
- 1:10:00the other one too. Mark and Colt AI does not only
- 1:10:05change the information, it changes our environment and
- 1:10:10which which people pay attention how they feel, trust again doubt
- 1:10:21and decide. That means AIAI affects the
- 1:10:33state in which people participate in democracy.
- 1:10:42So I will changing the quite a environment faster than human
- 1:10:57beings can psychologically adapt to it.
- 1:11:01That's the question. Yes,
- 1:11:18this is also very important. The next topic here, the danger
- 1:11:23is not only that people believe in something false, the deeper
- 1:11:31danger on that, on that all, it's that people lose the
- 1:11:37ability to believe in any shared reality at all.
- 1:11:42What is this what I'm looking at?
- 1:11:44I don't know. I don't care.
- 1:11:47So when everything can be fake, trust, not simply move, can
- 1:11:57collapse everywhere. So how can a democracy brief if
- 1:12:07people no longer share enough reality to what do you delegate
- 1:12:18together? Yeah, deliberate, very
- 1:12:26important. So yes,
- 1:12:43we should not evaluate AI high performance.
- 1:12:48The other ones was to talk about it too, about the speeds or
- 1:12:55intelligence. We should ask what kind of what
- 1:13:02kind of human? What kind of human?
- 1:13:10I lost Lion here. No.
- 1:13:21The question I will leave here for, for, for this one is what
- 1:13:27kind of human beings do our system leave behind?
- 1:13:33What are we going to be with this?
- 1:13:38Very important, yes, I brought it last time I was participating
- 1:13:49in this. I will.
- 1:13:55I will leave you something, but no, I can see you.
- 1:13:59I'm not well. I'm not here to map the whole
- 1:14:03high field. I possibly cannot do this by
- 1:14:05myself. So I'm here to ask for other
- 1:14:13human being inside democracy can still exist.
- 1:14:20No, not exist, not exist. Sorry, I lost it again.
- 1:14:32It happens. Well, I'm here to ask whether
- 1:14:43the human being inside the momentary can still breathe,
- 1:14:48think, doubt, trusts and response.
- 1:14:53Response. Not react, respond.
- 1:14:59The question is not only our well, maybe, maybe it's not so
- 1:15:13important in, in, in this this case, but but AI and democracy
- 1:15:20are not only about systems, about the platforms and
- 1:15:24regulation about this. It's it's they're also about the
- 1:15:29human being inside the democracy.
- 1:15:33You know, if you forget the human layer in the AI, we may
- 1:15:40build intelligent systems around a weakened democracy nervous
- 1:15:45system. I hope it's gives some meaning.
- 1:15:50Yeah. So I don't know all I want to, I
- 1:15:58want to to end this with a small exercise.
- 1:16:01You can see one minute of democracy, a small democratic
- 1:16:10act. Before you do something, before
- 1:16:13you share something, before you react, before you start working
- 1:16:18on something or anything else, take one minute, just one
- 1:16:24minute, 1 minute of the democratic act.
- 1:16:30Place your feet on the ground. Look away from the screen.
- 1:16:36A minute ask yourself, what am I doing right now?
- 1:16:43What is the purpose? Am I reacting or I'm responding?
- 1:16:51Is this something I don't see or that one one minute sound very
- 1:17:02small, Very small, maybe not important at all, but maybe
- 1:17:09democracy also live there in this moment, this one minute.
- 1:17:20And that's more pause before reaction becomes action.
- 1:17:29And by this I want to say thank you.
- 1:17:34Thank you so much, Justin. Thank you.
- 1:17:36I really appreciate it. I have to.
- 1:17:41Yeah, Thank you. Let's call all the others on the
- 1:17:44stage as well because we have the Q&A session.
- 1:17:49Hello again to everyone. Welcome.
- 1:17:53Thank you. Welcome.
- 1:17:57Well I I have gathered some questions from media for you
- 1:18:01guys. The 1st it's for used, it says.
- 1:18:06Which of the four democracy definitions do you personally
- 1:18:09find most convincing? Elections rights, equal equality
- 1:18:13or deliberation? Yours.
- 1:18:16Yeah. And, and I'll, I'll begin then
- 1:18:19by by answering a different question or a slight variant of
- 1:18:22that, because I mean, I, I enjoy much the, the different
- 1:18:25presentations that the Torsten and Paul and Mark gave it.
- 1:18:29But, but I would say that they lean all heavily into the
- 1:18:32viewpoint of democracy as the liberation.
- 1:18:37And that's a choice as, as Mark would say, and I guess a
- 1:18:44reasonable choice. But it also opens up, you know,
- 1:18:47the, the fact that there might be something there that's
- 1:18:50missing. I'm a I'm a classicist in the
- 1:18:54sense that, that I believe that democracy, you know, shouldn't
- 1:18:57be four wolves and a sheep voting on what's for dinner.
- 1:19:04So I mean, I lean over into rights and rights and
- 1:19:07liberation. That is the right of people to,
- 1:19:10to say no and also to say no to engage in in further dialogue.
- 1:19:17If we take it into sort of a more technical aspect, I think
- 1:19:22Mark's walkthrough of the of the different processes was, was, I
- 1:19:29think eye opening for me as well.
- 1:19:32I read his book before, but I think he he how it turns a great
- 1:19:35point. And the point here, Mark, that I
- 1:19:37think you, you, you helped me see is, is that there's this
- 1:19:42efficiency versus democracy trade off in a way.
- 1:19:48I guess that's what the different people in
- 1:19:50municipalities who employ AI, you know, must be thinking.
- 1:19:53Nobody sets out and says, you know, I want to use AI because
- 1:19:57then I can destroy democracy, right?
- 1:19:59Instead, what they're thinking here is a technology that allows
- 1:20:01us to do things cheaper and faster.
- 1:20:04Yeah. But if we need to build in
- 1:20:09democratic safeguards, then that becomes more costly.
- 1:20:15But then it comes back to the question of who's paying for it,
- 1:20:18right? And if we, you know, don't want
- 1:20:23to have a very costly process, should we even have it?
- 1:20:26Would the person here on the with the rights view of
- 1:20:29democracy say? I have another one for you.
- 1:20:34Used, it says. Are we actually seeing
- 1:20:38democratic decline, or are we simply becoming more aware of
- 1:20:42democracy's old weaknesses? Yeah, I think I would argue the
- 1:20:50the latter really. I mean, I think, you know, we're
- 1:20:53discussing here potential risks and democracy is always at risk.
- 1:20:59And I think we should, we should take great care to remember that
- 1:21:03that the system that all of us here in this call have taken for
- 1:21:07granted in our lifetime. Well, well, we are historically
- 1:21:14and globally the selected few. I mean, democracy is a rare air
- 1:21:19that not all humans get to share.
- 1:21:23And I think that that democracy, of course, you know, isn't
- 1:21:28perfect. There is, it's something that we
- 1:21:32need to reinvent and also defend against many different things.
- 1:21:39AI is potentially one of them, but I think that that
- 1:21:42technology, we have to remember this AI is, is not a person,
- 1:21:46it's it's a tool. And a tool can be used for good.
- 1:21:49It can be used for evil, and it can be used wisely, and it can
- 1:21:54be used stupidly. I have a question for Mark.
- 1:22:02Marcus says. When a public institution
- 1:22:06completely leases its cloud and data infrastructure to a full
- 1:22:10reign private vendor, how have they efficiently, effectively,
- 1:22:16sorry, outsource their core democratic sovereignty?
- 1:22:22Yeah, that's a great question. And it's something I've been
- 1:22:26dealing with extensively also on a or a chain.
- 1:22:30And it spills a bit over into course points as well, because
- 1:22:35what we're talking about here is digital sovereignty basically.
- 1:22:40And to say that just because we outsource certain parts of our
- 1:22:49sovereignty to a private vendor doesn't necessarily mean that we
- 1:22:54outsource our entire democracy. And a core was was talking about
- 1:22:58this as well, because you have to view it in a in a larger
- 1:23:02context. It is like cool stated, it is a
- 1:23:05stack of and and like I said in my presentation, it is a stack
- 1:23:09of different parts and and whether you sign over your
- 1:23:17digital sovereignty to a private vendor depends very much on all
- 1:23:20parts of this stack. So you can show you can, you can
- 1:23:24outsource certain parts without it necessarily being a problem
- 1:23:28exactly as as course. And I think that's, that's the
- 1:23:33good news here that we, we haven't signed over at this
- 1:23:36point our, our digital sovereignty in, in any sense of
- 1:23:42the word, but we are in serious risk of doing so because of
- 1:23:48conveniency. Basically, it's easier to just
- 1:23:53choose one vendor to deliver the whole thing.
- 1:23:56But if we're not careful about what vendor that is, then we're
- 1:24:01signing over potentially a lot of sovereignty to to the
- 1:24:06vendors. Maybe Cork can say some more
- 1:24:09about technical aspects of this, but that that would be my input
- 1:24:13and my answer. So we also have a question in
- 1:24:21the chat you wanna tip on, She said this this life is on the
- 1:24:28more interested of the all she didn't know.
- 1:24:31Sorry, I thought it's a question.
- 1:24:34But we take praise as well, don't we?
- 1:24:37Yeah. But Morella, if I can interject
- 1:24:41the small thing here, because Mark and myself, we've we've had
- 1:24:45the pleasure of discussing this before.
- 1:24:47And, and one of the things I like to point to when Mark talks
- 1:24:51about digital sovereignty, there's sovereignty, you know,
- 1:24:55as in Danish sovereignty, the Danish society, its sovereignty
- 1:24:59versus US versus other countries.
- 1:25:03But there's also the question of the individual person in
- 1:25:07Denmark, his or her sovereignty. That is, in a digital sense,
- 1:25:14when does a Danish citizen have the right to say no?
- 1:25:19I mean, should we, should we give, should you have a, a
- 1:25:23constitutional right in Denmark to say I will not allow AI to
- 1:25:28assist in the handling of my case because that is where
- 1:25:31sovereignty, if we take it down to the individual level, that is
- 1:25:34where I think it will take us. Well, I have a question also for
- 1:25:43how can an everyday user spot the moments where their search
- 1:25:47results are subtly shifting their cultural or political
- 1:25:51values? Thank you for that question,
- 1:25:55Mirela. I, I think I'll start out on, on
- 1:25:58continuing from from where I used to made left, but, but, but
- 1:26:02then moving on to, to, to this, because, I mean, when I did my
- 1:26:08presentation, it was the third out of four.
- 1:26:11And now that we have done all four, it's, it's much easier to
- 1:26:14connect all the dots. I, I was pointing in that
- 1:26:17direction where if we depend on Microsoft, it's an American
- 1:26:20company. But that's more to the story
- 1:26:23because from a Danish perspective, Microsoft has
- 1:26:26bought Danish companies. Microsoft have Danish employees.
- 1:26:29They have an office in Copenhagen.
- 1:26:31They have a, we, we can schedule a meeting there.
- 1:26:35They have data centers nearby. We can actually make agreements
- 1:26:39that that that we want them. And I think it's an important
- 1:26:42factor why many Danish organizations are still sticking
- 1:26:46to Microsoft. It might not be a Danish company
- 1:26:49as such, but it is slightly more Danish than Google or or or
- 1:26:55Apple or whatever they're called.
- 1:26:57And and it is actually possible to negotiate terms where you get
- 1:27:01your own data protected in your own way on your own terms
- 1:27:05because the alternative would be moving to totally different
- 1:27:09vendors. And that really connects you to
- 1:27:11your question middle. And sorry for for having this
- 1:27:15long path towards answering that because thank you.
- 1:27:20Because I mean, if you see that one source and if you get that
- 1:27:25from Google just as an example, that will be a fairly common
- 1:27:28example. It's not the only source of
- 1:27:30that. And you can prompt it in a
- 1:27:33better way. You can qualify it up to other
- 1:27:37sources. You can talk with actual humans
- 1:27:39instead of just using technology as so.
- 1:27:42So we still have options to choose from, fortunately.
- 1:27:46And I mean, if we want on the long term depend on American
- 1:27:51companies and if they move into at least a slightly European or
- 1:27:56even Danish perspective. And if we have alternatives and
- 1:28:00we do, I think at least on the short horizon, we're still good.
- 1:28:06But it does take decisions, it does take experience, it does
- 1:28:10take regulations, it does take a lot of things.
- 1:28:13I'm from a generation where a lot of things were originally
- 1:28:16not handled digitally and definitely not using AI.
- 1:28:19So I'm used to using those alternatives.
- 1:28:22A younger generation will start out from a different
- 1:28:25perspective. So I mean, on the short horizon
- 1:28:28I'm really not that worried, but I am very aware that it could
- 1:28:34drift in a direction that I wouldn't really prefer.
- 1:28:40I also have. Thank you.
- 1:28:42I also have a question for Chosen.
- 1:28:45Chosen says How does the total elimination of cognitive
- 1:28:50friction in seamless AI tools silently weaken our psychology,
- 1:28:56psychological resilience, and our capacity for critical
- 1:28:59thinking? I hope I have some more English
- 1:29:06words left. Should I repeat it chosen for
- 1:29:10you? Well, I think when when teams
- 1:29:13began to to trust the system more than their own shared
- 1:29:22judgement, Something, something important is lost, you know, in
- 1:29:29the in the danger. It's not only that AII makes
- 1:29:35mistakes that I think the deeper danger is that, yeah, the human
- 1:29:42has stopped practicing the the courage crush to question them.
- 1:29:51Yeah, OK. Sorry, sorry, sorry.
- 1:29:59Yeah, trust them. Yeah, that's what I want to say.
- 1:30:03I have another one for you. Tossing please.
- 1:30:06In an in an era defined by algorithm drift, how can
- 1:30:11individuals anchor their sense of identity when their daily
- 1:30:16information stream is hyper personalized and constantly
- 1:30:20mutating? Interesting, right?
- 1:30:26Yeah, let me think. How can be that they can stay
- 1:30:32grounded in that sense, I suppose this person asked.
- 1:30:36In this world of, you know, everything, it's hyper
- 1:30:40personalized because the algorithms quite often, all the
- 1:30:45time, actually constantly show us what we actually supposed to
- 1:30:50see without our being aware, very well aware of that
- 1:30:56algorithm. So in that sense, how they can
- 1:30:59actually stay in that state of being more grounded and more.
- 1:31:03Well, well, well, well, if the feet constantly adapts to us,
- 1:31:16you know, so also slowly adapt it also slowly adapters.
- 1:31:24So the anchor, you know, this, what you were asking about must
- 1:31:29be something outside this algorithm, you know, body
- 1:31:39memory, relationships, plays, values of conversation.
- 1:31:46This is that you can do. It's my.
- 1:31:50It's my. Well, I kind of finished my
- 1:31:59questions from social media. Does anyone of you has something
- 1:32:03else to add? Because we kind of achieved also
- 1:32:06the end part of this life event. Just a short comment from me.
- 1:32:15I'm so happy that we still have option to choose from.
- 1:32:20I'm so happy that in this live session we can do the long form
- 1:32:26dialogue about these topics. So many decisions are made in a
- 1:32:31rush and, and, and today we are taking our time to not answer in
- 1:32:38a rush to, to really put more perspective and really to
- 1:32:42supplement each other. I hope it will support a people
- 1:32:46who should make their decisions, whether it being individuals,
- 1:32:50whether it should be people in in power.
- 1:32:55I don't know if that hope is naive, but at least I I hope
- 1:32:58that we can do our part because I really, really, really want to
- 1:33:02protect our democracy. Thank you.
- 1:33:05Used to you would like to add something.
- 1:33:09How? How can you get?
- 1:33:10How could you guess, Marilla? Just looking at your face, my
- 1:33:13dear UST. As a professor, I always have a
- 1:33:18lot to say. But but look, I mean, I think,
- 1:33:21you know, that that we should be wary about this notion that we
- 1:33:27were living in a in a golden age of democracy where everything
- 1:33:30was settled, everything was good, and then AI showed up.
- 1:33:35Democracy is a fight, as I said before, that each generation has
- 1:33:39to fight and AI will help us in this fight.
- 1:33:43It can also harm us. It exaggerates, as Torsten
- 1:33:46pointed out so wonderfully, this human notion that whatever we
- 1:33:51see in front of us is all that there is.
- 1:33:56And the antidote, I guess, is, as with a lot of other questions
- 1:34:01in in AI, simply to remember that AI is just popping up a
- 1:34:06mathematically calculated, statistically, stochastically
- 1:34:10relevant answer in front of you. It doesn't have meaning because
- 1:34:15it can't have meaning. We humans can have meaning, but
- 1:34:17we're tricked, and we and the person that tricks us is
- 1:34:20ourselves. We end up tricking ourselves
- 1:34:23into ascribing meaning into what the system is showing us, and we
- 1:34:29simply have to be mindful of that.
- 1:34:31We can't stop ourselves from doing it because it is, for lack
- 1:34:35of a better term, of human nature.
- 1:34:38And on that note, I think that's, that's my summary of of
- 1:34:41what I think I said today. Yeah.
- 1:34:44Yes, Mark, please, you're very welcome to share.
- 1:34:47Your thoughts? I just think that one word
- 1:34:51popped into my mind is listening to both used and, and to core,
- 1:34:55and that is awareness. We need awareness.
- 1:34:59Awareness is the starting point. And, and I'm talking about
- 1:35:05awareness of what AI is and what it can do and what it should do.
- 1:35:12And, and, and this is the basic, this is the starting point for
- 1:35:17what used to saying about individual sovereignty.
- 1:35:23It starts with the awareness. It starts with the understanding
- 1:35:27of what are these systems. We don't have to have a deep
- 1:35:31technical knowledge, but we need to be able to make an informed
- 1:35:36decision. That is where it starts.
- 1:35:40Then awareness in in the institutions as well.
- 1:35:44And, and I think we are at a point where there are beginning
- 1:35:47to be an awareness about where do we lease out our sovereignty,
- 1:35:54how do we keep as much sovereignty for ourselves.
- 1:35:58So really I think awareness is the basis of all this and, and,
- 1:36:03and the basis and the starting point for everything we're
- 1:36:06talking about. Tausan, we're the.
- 1:36:10End of the day, it's all about humans, isn't it?
- 1:36:13Fortunately. Exactly.
- 1:36:18Tausan, would you like to add something so we can close this
- 1:36:21live session? Well, and all this talk about
- 1:36:24democracy, I've been out fighting for it for some while
- 1:36:30ago and worse. So this topic goes deep inside
- 1:36:36of me. So I definitely want to protect
- 1:36:41democracy and sometimes I try to fix it, but I can't do this by
- 1:36:47myself also. We also have to be like and call
- 1:36:53and everybody else. See now Mark Ross's aware
- 1:36:57awareness. You must be aware of what AI is.
- 1:37:02It's like a kind of, I'll call it elephant Slam, 2 point O, but
- 1:37:07you should be aware of what you wish for.
- 1:37:11That's it. Thank you.
- 1:37:14Well, thank you to all of you for attending this very, very
- 1:37:18strong and powerful episode, AI and Democracy and see you again
- 1:37:26after the summer break. Thank you, Chao.
- 1:37:32Chao.