Hello, I’m Nicholas Thompson. I’m CEO of The Atlantic. I’ll be the moderator today.

We are going to have a fabulous conversation on regulation of technology and AI and deregulation. My hope is that at the end we have a little bit of common ground, people with very different backgrounds, what laws should be changing, what needs to change structurally to get that done, and what happens next.

So let’s introduce this absolutely marvelous panel. Starting here on my left with Maryam bint Ahmed Al Hammadi. She is the Minister of State, Secretary General, UAE Cabinet.

Joel Kaplan, Chief Global Affairs Officer of MEDA. Yutaka Sasaki is at the end. I wrote it down.

How are you doing? The present CEO of NTT Data. And Federico Sturzenegger, who says you can pronounce his name just like his uncle Arnold, the Minister of Deregulation.

Best title on this whole panel. Kind of amazing. Are you the first Minister of Deregulation in Argentina?

Yes. And you know when we started thinking about this with President Millet, he said, well maybe you should be the Minister of Modernization. I said no, no, no, because the state doesn’t modernize.

The state pushes you backwards. So it would be, if we want to improve the state, that would be the backwardation minister. And I think that sounds very good, okay.

And also President Millet says, I also don’t want you to be the simplification minister because I don’t want you ever to think about simplifying something without asking yourself before if that thing that you were about to simplify should exist in the first place, okay.

So we kind of ask a deeper question and that’s how we got to the deregulation and the name.

That’s a cool, it’s a cool way to actually push policy and perception through titles. Yeah, I’m gonna start changing all the titles at the Atlantic. All right, so let’s start with Minister al-Hamadi.

So you have changed a remarkable number of laws in your country. Yeah. Believe it’s something like…

85% over the last 90% 90% I was out of date. I googled this three hours ago. So 90% of the laws Explain and it’s in the last three years

four years four years

explain the most Consequential and interesting changes that you’ve done for tech and for AI

There was a direction from the leadership in the UAE that we need to Relook to our laws when I’m saying about Regulations we have different layers of regulations. Now. I’m talking about the first layer, which is the law itself.

Yeah We have laws that is was valid since 30 years 40 years 50 years ahead So we will we need to know we actually did some type of deregulation Regulation in terms of that we look to our relations and To see which one of them is still valid and which one of them we need to repealed so there was Hundreds of teams working parallelly under the supervision of the cabinet in the UAE and All the teams they working because as you know United Arab Emirates is a federal country So we have Laws at the federal and whenever we do any law we have to coordinate with the federal entities and with the local Governments also we have seven local governments.

So actually the process was We were struggling a lot but everyone has the same vision because we has been cascaded from the top to down that we need to do that changes and We could within the four years 90% of our laws have been updated.

We did massive revolution in our regulatory framework The articles, even though in the United Arab Emirates, thousands of articles have been changed, thousands of them have been modified. And to do that one, actually, it took a lot of effort from us. And then we thought six months back, how can we use AI to do for us that revolutions?

So we don’t do it again as manual. And here there was a project that has been approved by the leaders in the UAE that we want to develop intelligence led regulation model. We need a model.

Is it somebody will tell me the first time, are you going to use AI tools to draft for you your laws? We said, no, we need more than this. We don’t want only, for example, Chad GBT to draft for us a law.

However, we need to have a model that listen and speak to the social media. So any law that we issue within the UAE, we need to listen at a larger scale to all our stakeholders to see what is their comment or what is their feedback about the law. So anything that, and then with the AI to analyze that one and to tell us that article.

The model, that law, needs, there is a lot of comments, positive or negative, about it. And then he will suggest for us what changes we have to do in that specific law. We need that model to just speak to the court.

So we want to assure that the law, when it goes to the court, is being implemented right. So the court ruling, is it compatible with the law or not? We need also a model that speaks to the service that is delivered to the customer and stakeholders and our investors, to see if our requirements that we put in the regulations, is it complicated?

It is too much? Is it less or more? So we combined all these features in the project that we are working with one of the AI company.

Yesterday we have launched the first white paper about what we have learned so far in this project. And we hope that within the project, it’s a two year project, we have a model that can be shared with all the government worldwide. They can share, they can learn from it.

And there will be a lot of case studies about it.

Fascinating. Interesting stuff. Lots in it that I want to get back to.

But let’s go to Minister Sturzenegger. You’ve knocked out, you haven’t changed 90% of your laws, but you’ve knocked out 13,500. And what was the most consequential and beneficial when it comes to AI?

Okay, let me see where to start. We started working two years before Millet became president. We reviewed all the laws of Argentina, and we classified them, laws that had to go, laws that were okay, and laws that changed.

And we had all these things prepared by the time he became president. And this is what allowed us to… who make such a swift reform in one year and a half.

So deregulation requires some preparation, correct? Now, we didn’t have an AI law. So I didn’t have any AI law to deregulate because there was nothing.

So my only task in this moment with AI law is that no AI law appears, okay? So we want in Argentina to make sure and give the message that we do not want to regulate AI.

So you added a regulation.

No, no, we don’t have, we don’t have. So I want to make, but all the congressmen, they all want to have a law regulating artificial intelligence.

But isn’t it a law saying we can’t have a law?

Oh, well, if you want to put it that way, yes. We don’t want to, we want to have, we don’t want to have anything, okay? Just call that as the way you want.

Then I think actually two other questions which I think are interesting in the relationship between regulation or deregulation and artificial intelligence is, will artificial intelligence change the need for regulation at large?

Let me give you one example. One of the reasons we regulate some sectors is something called asymmetric information. So for example, there’s asymmetric, a financial sector is regulated because you think there’s some asymmetric information, the depositor kind of doesn’t know about the bank, so you have like a regulated system which kind of provides the guarantee, et cetera.

So asymmetric information in many areas of economic activity, you have asymmetric information in many markets, is the justification for regulation. Now I say, if artificial intelligence gives us a lot of knowledge, doesn’t it solve itself, the asymmetric information problem? And if it solves the asymmetric information problem, then we don’t need to regulate that.

So I think there’s a relationship of artificial intelligence maybe changing the need, this is a question I’m just posing, I’m just proposing the debate, but I think it may be the case that it may eliminate the need for regulation in certain areas because it will help solve those issues that the regulation was supposed to solve.

So I think that’s an interesting line of thought. The third one is, can we use AI to generate the deregulation, which is a little bit what the minister was just mentioning, and we really haven’t gotten there yet, because the reason is we’re not so much in the business of rewriting regulations, but in the business of removing regulation, because sometimes we have this fantasy that regulation in an economy is built like by a well-intentioned central planner, you know, some bureaucrat thinking about how to do good to society.

Well, of course this may be different in different countries. In the case of Argentina, my finding is that most of the regulation is not the result of a benevolent central planner, but it’s the result of interest, kind of people using the state as a way of building a regulation, which at the end of the day is like a ring-fencing competition, generating privileges.

Deregulation also comes from interest too.

Well, I mean, if the regulation is the result of, which is what I find in Argentina, which regulation is the result of some sector which has been able to generate something which blocks entry, which is really what we’re going after, then in that case deregulation is really kind of an obvious choice.

Cool. You’re clearly the man in favor of deregulation. I’ve talked to you about it before.

Representative of the United States, our president, was just on stage talking about, what was it, 130 laws he cuts for every one he adds. But what are the laws that you like, right? I know that like Section 230 of the Communications Ceasancy Act, that law doesn’t exist.

Facebook’s got a lot of problems. Meta’s got a lot of problems. What are the other laws that you think are important to keep?

So first of all, I just want to say that what both of the ministers said was really fascinating to me because they’re examples of people who lead organizations who are already thinking. deeply about how AI is going to change either the way that they conduct the work of their organization or have implications for what the substance of the work is, which is what every organization over the next couple of years is going to have to do.

And the ones that succeed, whether they’re government agencies or whether they’re businesses, are the ones who start thinking now about how AI is going to change their workflows and change the nature of the work that they do.

So I really, I thought it was really gratifying to hear both ministers on that front. Is your question about regulation that we like on AI generally or on regulation generally?

Not like speed limits. I mean, yeah.

Well, there are some, you know, for the social media part of our business that we could talk about, but not here. So look, to power AI, you basically need four things. You need talent, you need data, you need compute, and you need energy.

So you have to have a regulatory environment that ensures access to all four of those things. And for the most part, what we’ve seen in the U.S. under President Trump’s AI Action Plan is removing barriers to innovation across those four key areas.

And I think that largely is what’s necessary to unleash the investment and the progress that we want to see on AI and that the President correctly views is necessary to win the battle, the most important national and economic security battle that we face right now, which is the AI race, and in particular against China.

So the AI Action Plan removes barriers to innovation across federal agencies. It ensures energy and data center permitting reform. So some of those reforms are positive laws in terms of what they require agencies to do in what time frame to make sure that you can actually get a…

the electricity generation plant that you need for a data center built. You know, you’ve got, you know, China built 440, or put into performance of 440 gigawatts of power in 2024, and the U.S. did about 50.

That’s a huge advantage that China has. We need to be much better about that. So we need to put in place regulatory structures around energy and transmission grid permitting.

That’s the kind, those are the kind of laws that I think are very positive. We have to have laws, I know this isn’t necessarily popular at the Atlantic, but we have to have laws around copyright that ensure access to the training data.

I like laws around copyright. Oh, about access.

Fair use.

And we believe it falls on the other side, so I’m not mistaken.

Ensuring access to data for the training. The models depend on having access to huge pools of data. And any country that doesn’t ensure that that’s going to be the case is going to be left behind because their data is not going to be included in the training models.

And so, you know, if you want, you know, if you’re another country that’s worried about AI sovereignty, really what you should be worried about is making sure that the large language models that are developed include data from your country so that the models reflect the culture and interests and expertise from your people.

So those are a couple of examples of areas where I think, you know, you do have to have kind of guardrail, but really the guardrails that you put in place are to make sure that you have access to the things you need to power these data centers.

Great. I’m all in favor of regulations that respect the licensing deal that you and I are going to strike right afterwards. Yutaka, will you respond?

You’re a business leader. You operate in all these countries. You operate in 50 different countries?

Yeah. Yes. Yeah.

Over 50 countries.

So actually, let me ask you this way. How do you feel when you hear Argentina’s deregulating everything? Because what you want is you want clarity and similarity across regulations, right?

Explain how you felt listening to all this.

Yes, our position is that we accelerate the business with using AI, but we need to align with the regulation. And today we have two government side people and two business side people. Our position is very, very business side.

And I think we need a trusted environment by the people, by the employees in business use cases. And I think we are not only the IT service provider, but also the data center provider. And data center, it’s so easy to be regulated.

It’s a very, a lot of rules and something related to the power supplied, yes, of some environment issues. And cloud layer, LLM layer and applications, they will be so very difficult to be managed by the regulation. But in the business use cases, we need to establish trusted infrastructure.

And that’s one of the key words is sovereignty. From the government view, from the big companies view, we need to manage the confidential data that they are reluctant to store their confidential data in the public AI. And so we need to manage their confidential data in the sovereign cloud or sovereign AI.

In AI case, we’re calling the private AI. Of course, we need to use the public AI, very intelligent employees that would like to use public AI. But in the future, we need to establish private AI.

And in the future, we will manage hybrid AI, the combination between public AI and private AI. That will be the very trusted environment.

But that makes very good sense. Minister Al Hammadi, will you talk a little bit more? You mentioned this public feedback, which I think when I was listening, this is one of the most important issues because one of the hardest things with AI, certainly in my country, is how little people trust it.

Like they don’t. I love AI. I think it’s amazing.

I use it all the time. It’s awesome. And like most people hate it, right?

So you need, or not most, a majority. You’re building trust in part by having feedback on everything you do and building using AI to help get comments. Explain a little more about that, because it’s fascinating.

Can I zoom out a little bit?

You can zoom out. You can zoom in. You can pivot sideways, however you want to answer it.

OK, because that is only one angle. When I spoke about the model that we are developing in the UAE, there is, you know, with the AI, always you think that we will have more agile, faster regulations. But actually, there are some of the principles that we put in the model that we are developing that we cannot compromise.

Which is the rule of law foundation and constitutional safeguards. Let me explain. For example, equity before law.

We have we are. We are not allowing that FAST to become a bias. So whenever the model or the AI tool is giving us any biased or harmful outcomes, we have developed in our model a mechanism to stop it.

There is also the rule of law. This is one of the principles that we put in the AI model. We need all the AI output to be traceable to a legal basis, not only to statistical patterns.

There is also fair procedures. AI can detect for you gaps, can tell you about the recommendations, can give you outcomes, can analyze, can tell you the red flag risk, but it will never bypass the procedures. I will tell you, for example, AI can tell us the non-compliances, but it will not impose penalties on the community.

So this one is not allowed. It’s one of the principles that we use in developing the AI model. We have also, as the professor was saying, we have put the principle in the model that we have regarding privacy and data protection.

So we need to ensure that one, because as I said before, we are dealing with a massive number of information. It’s related to federal, local, court, economical, social. We have many data.

We have to ensure that we need to tackle with them in terms of quality. We need to ensure that the data we have is quality, and it is consistent, because if we don’t have consistent, we will not have a good AI model that will help the government. So we need to lay out.

And this is what I have seen, is that there are many people who have been using AI at a large scale, in a very short time. And here is the feedback of the stakeholder and our investors. We can listen to them at the large scale using AI tools.

And also, there is one of the principles that we have put in the model, is the human accountability. Still we believe that at this stage, AI can advise, but still, human is in command. And this is very important for us at this stage.

Great. Minister, you wanted to respond?

Yeah, I want to touch on the question you just made. First of all, a quick remark on the copyrights. It’s a kind of an interesting legal issue, because we, for example, buy a book and we pay the copyright for it, OK?

But if an AI reads the book, it becomes the book. I mean, it can replicate the book perfectly. So I think, how do you solve the problem of property rights, that once you read it once, you become the book with perfection?

Anyway, it’s a question there.

Who knows? We’ve sued some AI companies, and the courts will decide.

We’ll see. We’ll see, exactly. But look at this.

Look at this. Alejandro, parate. Parate.

Sentate. Parate de nuevo. OK?

Sentate. So now imagine that he’s a robot, and he’s obeying my orders, correct? And so in a few years, we’re going to have people going around like that, and they’re going to be robots.

So we’ll have to build-

We don’t know, in fact.

We don’t know if we don’t have some here. He’s human. He’s human, by the way, OK?

He was just being friendly with me. So we have to build a legal framework. framework for these guys?

Who’s responsible for the actions? Because the robot goes out and goes out stealing. Who’s going to be responsible?

Okay, so we need to build a framework. So it’s an interesting, I don’t think it’s a very complicated, I mean, there’s, someone’s going to build a robot and then it’s going to sell it to someone, and then someone will take the responsibility.

I guess that’s it. But I think something that happens is interesting, is that in regulation theory, you have something called saliency. Or I call it the Baltimore effect.

Remember this boat that crashed the bridge in Baltimore a few months ago? Well, the boat crashes the bridge, pulls down the bridge, and then you change all the regulation on navigation. And there were 1 million boats that were going, no one crashed a bridge.

So the saliency is the politician reacts, feels that they have to react to a salient event, and then regulate and imposing a tremendous cost. And I say people make mistakes all the time. But I think that if Alejandro were a robot and he would make a mistake, people would immediately jump and say, oh, we have to regulate the robot so that this doesn’t happen again.

So I think also, I think…

So is part of your job anticipating future demands for regulation and preempting?

No. Well, I think that’s a trap. Because most of the, most of the regulators, they, they are very imaginative of all the terrible things that will happen to people if they’re free.

And I tell you, the imagination is, I mean, you can be the, it’s better than the best novelist in the world. No, I think we have to be in that sense, we have to take more risk. I mean, we have to see if something left free generates a problem.

And then react to the problem, but not try to solve a problem that we didn’t even know that we have. So I think we, but I think this is important because this question will come over and over again as AI starts to move into different activities. And I want to finish with one thing.

I’m sure you’d appreciate what I’m going to say now, which is the latest with the resistance to AI. Okay. So what’s the, so the last, you know, there’s this book by Isaac Asimov.

called iRobot. In the last, it’s a series of stories on kind of robotics and artificial intelligence written in 1960s. I mean, you read this thing today, and it kind of blows your mind away.

So in the last story, it’s about a political campaign. And there’s a guy who’s a very good politician. But people are suspicious he’s a robot.

So you have a debate in society whether we should allow. And then the politician says, no, we can’t allow having a robot to be in a politician. And other people say, no, wait a minute.

Maybe it’s better, you know? These guys are fair. These guys work 724.

They’re honest. So I think, but you see, I’m seeing artificial intelligence coming into lots of mechanisms. And I think we’re going to have a lot of resistance from the people who somehow are left to the sides because of artificial intelligence.

We may not have industrial jobs in 10, 15 years, correct? And so I think this is, we have to be aware that we need to resist this fight against the implementation of AI. Because if we don’t, we would still be using candles, OK?

Because the candle producer would have protested because Edison invented the light bulb. And we would not have implemented it. But fight against it.

Maybe, Joel, I would imagine that you agree philosophically here. But let me ask you, so some of the regulations we just heard about, like, let’s have a right to privacy for citizens in the UAE. And let’s build that into the AI we have.

Let’s make sure there are some protections against harms. Are there not a set of regulations that can build trust so that more people use it? And actually, the benefits spread throughout society?

It may be that there are new regulations that are necessary as the technology develops. I think the risk is that, and we’ve got some real world examples of this, the risk is that imaginative policymakers are much more focused on the risks. And the potential harms and much less focused on the benefits of innovation and the benefits to the economy and to growth I think we’ve the model we’ve seen with the clearest model.

We’ve seen of that is in the European AI Act the EU AI Act You had that there was work underway for a long time on the regulation the regulation was was technology neutral focused on real-world harms and then chat GTP came out and And immediately the EU Changed the focus of the regulation and began actually directing the regulation at the technology And and as a result, you’ve got you had a piece of legislation passed in the EU That was very harmful to innovation that I think really risks the EU Falling quite far behind in this new technological revolution in much the same way that unfortunately They’ve become less competitive over the last 40 or 50 years because of regulation and it was largely because they were they were very focused in those early moments on all of the possible harms that could come from the technology itself, so I think Waiting to see whether you know What kind of robots are developed and which which harms they create possible harms they create and then deciding do you need to?

Regulate against those harms and figure out who’s gonna be responsible for it and legislate that way. I think is the right approach Rather than just regulating the technology on the basis of all of the possible imaginary Harms that you might imagine you talk to you the same analysis of the EU AI Act.

Yes Yeah we understood you side that they have hard low and The compared it to the EU the United States side a soft low in Japan also that we and we have a soft low and we need to align the region by region and I mean As I mentioned, private AI or regional AI, we will have a lot of AI models in the world in the future.

I believe the edging AI model is different from the conventional IT systems. Conventional, traditional IT systems, it’s very programmed precisely. The input and the output is fixed.

But AI models, you know, the prompting and sometimes various answers are written. And I think the conventional IT system, it was easy to be managed. We can manage the traditional IT systems.

But AI is different. It’s very difficult. We need to monitor the performance and the quality.

And that is very important. And managed services will be very necessary in the future, yeah.

Minister Sursnegar, so one of the things that you mentioned is that, or not that you mentioned, but that you’ve said in the past, is that concentrated markets are good and that you should let companies get as big as the market will let them get.

Are there any areas where you believe there should be competition policy to try to prevent, say, you know, a company that controls one side of the market from using that power to get dominance in another?

No, we think there should be competition policies, but competition policies should be focused on entry. It should not be, it should be focused, I’ll give you an example. For example, in European markets, you punish someone if they charge kind of way above the market prices.

Yeah.

And they’re a dominant firm. Imagine an airline that flies a certain route and they’re the only guys, they’re dominant because they’re the only one, they have 100% of the market. And they charge an exorbitant fee.

are no competition problem there. Because anybody can enter into that route and fly that route. In fact, you want a high price to entice other people to come into the market.

And there’s a paradoxical. Actually we wrote an open ending, The Economist with President Millet on exactly this question that you’re asking, which came out last week. There’s a paradox, which is, most of the time, the restrictions to competition come from regulation.

Because regulation kind of erects all these barriers to entry, which make it more difficult. So I think you need to have, of course, if you have firms colluding to exclude other players, that certainly is like the U.S. antitrust approach to, I think that’s perfectly and should be applied.

But I think we don’t put too much attention to the need to make sure that the government is not itself responsible for generating the barriers to entry, which are responsible for, I think, markets which work less.

I’ll give you an example. Nokia, Blackberry, iPhone. So imagine that at some point we said, look, Blackberry has too large a share of the market and we should kind of split it out and split it into parts, et cetera.

I think it would have been a terrible mistake. We didn’t actually have a competition problem because a third player could come and compete and challenge that market. We call it, in economic theory, we call it contestable markets.

So as long as you have a contestable market, if you have an increase in returns to scale, which reduces the cost, well, as a society, we want to produce at a lower cost. So you want to profit from that, but always making sure that, for example, these guys, that he can be challenged by anybody. But to the extent that no one challenges, you have to let them run their increasing returns to scale and their efficiency.

So that’s kind of our approach to the thing.

To be clear, we feel like we have very fierce competition.

Oh, yeah, me, we. As I read in the government’s complaint, me, we, and. Let’s talk a little bit about the U.S.-China competition, Joel, because it’s quite interesting.

Given the AI competition, which is one of the most interesting stories in the world right now between the United States companies and Chinese companies, how should the government take that into account when setting regulations?

China clearly has an entirely different regulatory framework, and it’s not one that’s fully deregulated. Explain what U.S. administrations should do to make sure that America is maximally successful.

Yeah, I mean, I don’t want to sound too much like a broken record here, but I mean, I think the AI action plan was designed with exactly that.

And also add, I think another element that is so important is open source, which is the area where China…

Yeah.

Let me rephrase the question. One of the most interesting critiques of American regulation is that it made open source AI much harder, and in fact, ceded that to China. What could have been done differently to prevent that outcome?

So I actually don’t think that ultimately happened. I think there was a real risk that that was about to happen. We actually…

I mean, the first real powerful open source large language model that was released was ours.

Yeah. It was the framework for DeepSeek.

Yeah. Yeah. And it saw wide scale adoption.

We released an open source model called Llama, and the Llama models have been downloaded 1.2 billion times. So that’s an incredibly democratizing effort, right? Because that means that developers all over the world, academics, universities, governments all over the world have access to the output of these incredible investments that Meta made, right?

It’s very capital intensive, as you know, to build one of these large language models. If we open source it, then everybody has access to it for free. That’s what it means.

So during the Biden administration, that was really… We released Llama. I think in 2023 timeframe, or early 2024.

There was a lot of debate and discussion within the Biden administration as to whether open source was too risky, good thing, bad thing. We believe for any number of reasons, including the democratization, diffusing the benefits of AI across populations and regions. And from a national security standpoint, we thought that there was a real benefit to having the global standard for building on top of AI to be set by a US company with US values embedded.

The Trump administration ultimately embraced that position, and that’s reflected in the AI action plan. So the debate about whether open source is a net positive, I think has been resolved in the United States in favor of open source, and I think that’s a good thing. Now, the Chinese, whatever it was a year or so ago, DeepSea came out with their open source model.

It was quite good and quite competitive, and it’s achieved a lot of adoption since then, as have a couple other Chinese models. So there’s a real, I think, battle still for where the global standard is gonna be, whose technology the global standard is gonna be based on. But I think, fortunately, that was an area where there was a real risk of the government really hindering the spread of Western values in the technology, but I don’t actually think at the end of the day that that was why the Chinese ended up having some good entrance in the open source.

Good entrance and more downloads on Hugging Face than American companies. Minister, when you listen to the minister from Argentina, and you listen to him taking a chainsaw to regulations, do you wish you could do some of that too? How do you respond to him?

Do you think, great job, let’s do more of that, or do you feel like, eh, maybe that works for him and I’ve got something else that works for me?

I will be with him tomorrow at 8 o’clock I want to ask him a question, straight forward In Argentina, is there any regulation?

Is there any regulation? Yes, yes, there is a lot of regulation

How many?

I still have a lot to catch

I think, I feel that whatever is being done for example in the UAE or Argentina is very close because it’s only the name, His Excellency using the word of deregulations and the aim of us whenever it was four years back that we want to eliminate any law that is not valid for us at this time maybe it was good 20 years back or 30 years back but now it is maybe not valid for us so actually we did deregulations when we look to any law we look to the clauses of that law and we try to deregulate the requirement which is unnecessary and we try to make it compatible of the changes that happening in different sectors and that’s why I feel that any country that is working into laws or regulations, reforms actually they are doing deregulations at the heart or at the center of the reform they are doing One thing that also Mr.

Kaplan was saying is the people I feel with the AI or with the age of the AI government should not resist the change now actually we have to invest in our people and that’s why in the UAE we are now preparing the new generation of the legal professionals We need our legal to be blended between the law and technology.

We need to have regulatory data scientists that can handle the data, understand the data, and from the data, they can know how the law is being performed in the real. We need to have regulatory knowledge engineers who can convert the complex law text to something knowledgeable, something people can understand it. And this is actually what we are doing now, so we can have a system that we can work on it.

And I think our professional legal can also handle it, and this is actually what should we do.

Can I get a two-hander? Can I get a two-hander?

Yeah. You can counter.

No, I’m saying, I mean, the process of the regulations is different in every country. I think in the case of Argentina, it’s so extreme because we build a status quo society where interest groups have captured the state and have built a regulatory framework which was building privileges for them.

And I’m sure that it’s totally different in her setup. So I think we can be, we have to be much more aggressive in deregulation because much more of the law have this kind of, the original incentive, objective was not really the one that you want for a society which is more just and growth enhancing.

I just wanted to make that clear because if not, it seems that kind of I’m proposing the deregulation forever. I think every context is different. I just wanted to clarify.

We need to understand, so we need to have various regulations in the, aligned with each country, each region, the culture, some rules. You mentioned that we need to think about the relationship between regulation and innovation. If there is strict regulation, it sometimes prevents the speed of innovation.

And so we need to have a balance. And innovation will give the benefit for the people. And so too strict regulation.

We need to see the worldwide use cases. For example, it’s a very simple use case. COVID-19, there was various responses in each country.

And we studied which country’s use case was the best. I think AI technologies is very evolving rapidly. And the common understanding will be very important.

And the technology, the literacy. The regulator side and technology innovator side, we need to have the common understanding for the latest technology of AI. How do you think?

Let me ask a last question. We have one minute left and I can’t resist the temptation. But Minister Scherzenegger, we talked about how we’re going to regulate robots.

If there was an AI that became much more capable than we have right now. And it could really take into account public opinion. And it could listen to all sides.

And like the UAE’s AI, it understood the law. And was so much better than the president. Would you cede decision making to this AI?

Well, through a democratic process.

I mean…

What are you asking me to… No, I mean through a democratic process.

But it would be hard for it to be elected. I mean, I don’t know what the constitutional rules are in Argentina. But if you had this on your desk, would you say, tell me what to do, I’ll follow it?

I can’t do it as an official, I can’t take opinions, but we have a democracy and that’s the way we elect our rulers. In the story of Asimov, it was a democracy and it was an election, so people at the end decided, well, let people decide if they want to choose a robot or not. I won’t tell you how the story ends.

All right, thank you very much for your time, that was a fabulous conversation. Thank you so much, all of you, it was wonderful to be out of Sheristan.