- Reset + PDFPrint

President Kersti Kaljulaid at Chatham House in London

27.03.2018

Thank you for these kind words about me, myself and Estonia.

Indeed, digital is not the first big wave of legally permissive environment creation, which has brought investment to our country. I would think the first one was actually in the early 1990s when our cost levels were everyone's dream and you could easily attract money to your country by having a much easier tax system. We did it and later it was copied a lot. The next time, was indeed digital, but not only digital. At the turn of the century we created a permissive environment for population-level genome investigations, and now we are able (with the cost of hopefully not more than 25 euros per capita) to provide quite soon for 10% of our population information on how easy it is for them to get diabetes type 2 and other common genetically hereditary diseases. This demonstrates that if you do not have money but you want to provide your people with services, there is a way to do it, and this is to create a digitally permissive environment. Right now, we are indeed already thinking about how to regulate artificial intelligence.  Even if we know that we are very far from creating artificial intelligence, probably further away than Elon Musk is thinking, we do have lots of automated systems and regulating for one will also cover the other, the liability issues, etc. So we are thinking of how to make sure that this wave of technology will not pass Estonia by. For example, our Traffic Code can regulate for a situation of a car and a robot having an accident, and we have already had such an accident with a package delivery robot and a car. The car driver was found guilty. It shows you that it is more general. It is not just that we happened on a digital gold mine, we do it systematically in Estonia.

Seventeen years ago, we wanted to provide our people with digital services like Industry 4.0. We thought that if we would automatize processes and remove people from the chain of providing the services and delivering goods, then we could afford more, with our small workforce in the public sector and low tax burden of the GDP, which has never exceeded 35%. At the same time, we have a population that is looking toward Scandinavia for public services to be at a good level. Now all of our people count on online as a part of everyday life. Last year we had a hiccup because of a technology provider, some people had to go to a government office to restart their digital identities, and we almost had a riot. People had to wait for an hour at a government office, shock and horror. This is when we realized that societal disruption is complete. We have digitally disrupted the society. It is not any more "digital with if needed paper alternative", it is now "digital, which needs digital alternatives". Luckily, we had several ways of digital identification so we could continue with digital Estonia.

So presently we have a digital disruption, but I think you have it too. Curiously it has passed most governments, but if you think of your people and your businesses, there is not much difference between an Estonian, or an English or a French person. We all partially start and end our transactions online. I am often asked why people trust banks online but not, for example, university applications, which is a government service. Here I have no answer, because governments really need to think seriously: why is it that their people do not trust their services? Why are they perfectly at ease when the other partner is private and deals with, let us say, their money. We cannot answer why our people trust our government provision of service. Part of it is that, we have proven to people that you cannot nose around databases and systems. It is an offence if you do it, and you will be punished. It is not OK if you are a police officer and you have in principle technical access and you can check your ex-boyfriend's new girlfriend's salary level or whatever in the system. It is an offence and you will be taken to court for it. We have had a few cases of that and people have learned very quickly not to do it. So we have demonstrated that the government actually does protect people's data. If Estonian police checks my tax record, I will be notified.

Here comes my second puzzle that I cannot solve for people of other countries. I am constantly asking myself: why people do not trust an e-health database, which will warn me if someone checks my data but they do trust the doctor's office where you do not know who last read and made copies of your data.

Very often the problem with digital is, that we ask for absolute security, when we should be simply looking for a comparison with the analogue-world security. There digital normally trumps it. This might explain why Estonians trust and others have difficulties with trust.

We have something, which is vital for the digital disruption of a society: the public and private sector all build on the same platform, which is called X-road and we use the same system for all services.  This makes Internet inherently safer for users. First of all, we are not anonymous online, we have also online passports—our digital ID. So I know the other partner with whom I am talking, and in addition the channel we use for talking is encrypted so no one can listen in. I think this is extremely important that we provide this kind of security that people know whom they are talking with. On the other hand, it also means that since the only thing we provide is this communication platform for different parties, the number of services can go up independently from the government; services can be offered by private companies as well. Nowadays we also have the e-residency offer for people who are not Estonian citizens but who can join our platform from the outside and use some services, which are provided for them and create new services. This also enlarges our sandbox for new technology testing.

It is as important to stress that at no point in time have we created any cutting-edge technology. Estonian digital identity technology was used by private companies for a long time and we did not invent it. For example the Finns had it before us, we actually knew they had it and we used it as well. If you think about how much the Moon landing changed your everyday life, I do not know how much it did but I am sure the washing machine helped much more. Low-level technology in the hands of the whole population with high penetration rates is much more beneficial than cutting edge in the hands of 1% of the population. This is another lesson of the Estonian ID and digital identity system. It is wholly inclusive. We achieved it by simply sneaking the digital ID on our travel document for the EU – so everybody had it and could use it.

The first service we offered was a traditional one. There is a saying that nobody wants to see their taxman, so the Estonian online tax board is the first online service we offered. In March 2000 physical persons got the possibility to file their electronic income tax returns via Internet banks for the first time. By 2003 63% of VAT tax returns and 53% of income and social security tax returns were submitted by electronic means, because people are inherently lazy and if they realize it is safe and easy they use it. Now it is VAT returns 99.3% and natural personal income tax returns 95%, so everybody is doing this online.

What has the state gained through this in addition to customer satisfaction? Looking narrowly at the tax collection, then in 2004 we spent one euro to collect 100 euros and in 2013 0.4 to collect 100. So the reduction in the number of employees is quite comparable – 44% of the workforce released. Luckily Estonia has always been a low unemployment environment so it has not hurt us, instead it has freed people for more productive work.

Coming away from the tax discussion to the more macro level, then nowadays we save 2% of our GDP only by signing digitally, not counting all other services and digital development. I lived outside Estonia in a very developed country in Europe, and compared to my personal administrative load, which I had to carry compared to Estonia, I think we also save around 3-4 working days of everybody's time. This new environment in the digitally disrupted society is also economically much more effective.

I promised also to look around the corners and to the next challenges of digital society. Since in Estonia we are already around the first bend, we may see a little bit clearer what is awaiting us and the new questions. Before I go into this, I think it is particularly interesting that it is today that we are talking about this, because the whole world is standing in awe and discovering what can go wrong in digital. We have always known that giving our personal information means we get free services and someone needs to make revenue off of it. Now that it is exposed, we say in shock and horror that we thought it is about Nike and Adidas, not about political parties.

It is very important we do not confuse things. If you have a safe environment where people can identify each other, then digital channels technically are not bad in any way. The carrier, the technology, is never the guilty party. It is only how we regulate or do not regulate, or what we understand or do not understand. It all comes down to identification. Some people really do not like big internet companies, but as soon as they need to talk to businesses, what do they do? In Estonia, I use my state-provided signature; most people rely on Amazon or Google identification tools, without any state guarantee. I think it is a bit irresponsible for governments to leave their people and businesses alone in the digital sphere.

Anyway, if I think about what is to come, then while the EU is grappling with taxation in the analogue world and trying to understand who gets the tax revenue of a Starbucks coffee delivered somewhere in Europe, we are already seeing a need to totally rethink our tax system. For that matter, also our social services delivery system because the jobs are changing because of technology. This is something I like to call the Alice in Wonderland problem. You know, when the cat was disappearing and the grin still stayed on. If the industrial jobs disappear, then what about our social model, which is based on the tax river coming in from big companies where people have gathered to work more efficiently? It has been an easy model for collecting taxes because people work in one company, they have contracts for life, or for a certain period of time, and it is all very organized and simple to understand and follow. I do not think that technology will allow us to stay with that solution. First of all, I think that as we saw in the agricultural sector, jobs will disappear. We provide our food with 3-5% of the workforce in the modern world and I think in industry we will similarly soon see that we will provide all the goods we need with a similar proportion of workforce, so the rest will be doing something else. This will be very different.

What it will be, we do not exactly know, but we have some hints. First of all, we know it is not going to be terribly high-tech.  For a long time it has been thought that the vulnerable classes were those, who were working blue-collar jobs. I think it was Jerry Kaplan at Stanford who mentioned that automation is blind to the colour of your collar. As with the example I gave about Estonia and the taxman, it is indeed true. The new job creation is more democratic than we might think, because you can be a travelling YouTuber making money out of this, and it does not take a degree.

For example, in Estonia somewhere in Mooste in a southern Estonian county of a couple thousand people there lives a man from South Africa, who makes bows and arrows. And they are world-class and his market is global. So probably none of his customers live closer to him than 1000 km. All these handicraft jobs are liberated from the need, not only to travel around markets in Europe, but also trying to make agreements with souvenir shops. So you see, new job opportunities are actually being created. Some of them can be quite specialized and can help those people, who have difficulties in living in our socially very demanding world. For example, an autistic person who agrees to knit only red socks can nowadays make a living, because probably globally among the billions online, he will find enough online takers for his red socks and he does not need to talk to anybody in the process. I do not agree with those, who say that these new jobs will only be for PhDs and narrowly selected people; I think the change is actually quite democratic.

If we think back to the industrial transition, it was painful for people who had lower financial cushions. Why was it painful? Because there was no social system and no education. Now, we need to keep on providing these to people while our tax base is being eroded. Because jobs are changing and we need to completely rethink how we gather our taxes.

Take this example: I have an MBA, I am a financial manager by education. In the morning, I work for someone here in London and then in the afternoon I work for someone in Australia. I choose to live three months of the year in Tallinn and six months a year in Sydney and somewhere else, so...globally. I work globally. I have a narrow set of specialized skills which I can sell globally. I do not need an enterprise anymore. I can sell directly. This is already happening here. The number of self-employed people is growing in the UK and the Netherlands and elsewhere in the EU. So, narrow specialization and different work in different places. Who will get my taxes and how? Who will provide my social services? I want my children to have an Estonian language education and I want it online. Living in Australia. This is how we need to think about how digital disruption is really changing our societies.

This is not sci-fi. This is already happening. If you look at the macroeconomic statistics, numbers of self-employed people are rising, the basis on which people make their living Airbnb and all that, is widening. The understanding of work as being something 9 to 5 in one country in one company, with a fixed home address and a fixed job address, it is going away. However, it is good, because people have more freedom, they have a better work-life balance. It is easier for women to manage babies in high-demand jobs in this way. So it is not bad. It is good. But we need to understand that this is happening and we need to discuss not simply Starbucks, but also need to think about the digital environment.

I think this will trickle down to all of society. Much more people will work independently and the education system needs to adapt as well. One thing we know for certain is that going once through the pain of getting a master's degree or PhD will not take you to retirement anymore in this quick technological change. You need to therefore also radically change how you teach people. We already see universities responding to this challenge of shorter-term courses that do not give you a degree, but give you a skillset. I think we need to observe our kids, see what they are doing and achieving before the end of the school system, and think if we can really afford to continue educating these kids the way we have been doing. The children in the kindergarten try to enlarge the window panels to make their mom come closer so their mom will come from the door. This is a new skillset we need to teach people, the difference between digital and analogue. At the same time these kids, if they are not English, already learn a second language before they enter school, because YouTubers mostly speak English. I have a child not only fluent in French and Estonian. His smaller brother was fluent in English as well, not because of the internet, but kindergarten. He realized he has a deficiency in his education. It took him a year. He is now as fluent in English as I am. No lessons, nothing. OK, he was French- speaking, so the vocabulary is the same, only the grammar and pronunciation is different. It is easier to come from French to English. However, kids know a lot more than we did, and they know different things than we did, yet they go to school the same and study more or less the same curriculum as we did. We also have an artificial border to people entering different education systems. When I was 15, I wanted to know why I cannot listen to university courses, if I am able to understand them. The problem is still there, I think. We need to make all kinds of learning levels available to people, never mind their previous educational experience, and never mind their age. Particularly the age, because as I said, no one – and I always say it when I speak to students graduating from university – no one here in this room today will still be making money from the same things in 30 years' time. Doctors probably have the best chance. Even there, think of e-health development. Your national medical system – and you are so proud of it here and rightly so –  they are testing diverting calls which they have identified as not being life-threatening to an automated system that will help you to diagnose and then take the right decision where to go.

Looking at people's behaviour online—what do they do, if someone is ill? They usually Google. I think it is our obligation to make sure they find factual machines to help them take the right decision whether to call the doctor or not. Right now, mostly what they find is voodoo and weird alternative stuff. I think we need to understand that also this profession, which is the most stable, is changing.

We have a lot of things, which we need to do, and I have not even come to artificial intelligence. I know that Elon Musk says there is a 10% of it by 2025, we will have the Singularity and a 50% chance by 2050. My first education from university was genetic engineering. I have to say, when I was studying, everyone was sure by 2010 we would clone a dog and by 2020, we would clone human beings with no problems. We all know that has not at happened so it might be with AI. If I read the full scientific analyses of the apocalyptic future, I take them with a pinch of salt because it might not ever realize.

However, it does not matter because already now there are systems, which can learn to the extent that they seem intelligent to us now and we need to think what to do with them. I very much like the example of a security question – for example we know in the history there has been a case of someone sent a worm to somebody's nuclear reactor to make it less useful for bad use. What if you send this worm now and it has learned quite a lot and is able to make decisions – not independently, but automatically. It is not singular, it is not thinking, but it is analysing the data it has. You think you can control and predict the outcome because it is not a singular AI, but you know it is smart enough to use the data it is finding and you think it finds certain information in this nuclear system.

Let us take now another system element, and this is someone who has used a contaminated computer in the system. The same computer went online and read, for example news from last week and the traces from the news are now in the system. So this AI or seemingly AI, very clever automatic worm, sees this information and knows it is in the nuclear system and it knows it is military. The news that it is reading is that the UN is planning to ban artificial intelligence use for military use. What will it do? It does not have to be singular – it can be devastating. Some people say AI has no reason to act unreasonably when we have finally created it or something similar. It will not turn our world into paper clips, which I believe is the most common example used. That it will use all our resources to make the world into paper clips, because it got the wrong signal. And this will not happen, because such a stupid signal will not be acceptable for an AI.

Now I am asking what are we currently doing? Today, and less than a couple months ago, we are using real resources – electricity, cooling, brainpower, computing power, to do what? – to mine bitcoin. Is bitcoin real money linked to any real economy? I do not think so. Did we know something would definitely go wrong with it? Yes, we definitely did. Did it keep us from wasting resources, turning huge amounts of resources into bitcoin? It did not. So, I mean, we can do it, and we are fully singular by the definition of AI, and why would the AI not do it?

What we are trying to do in Estonia is to pilot in 2018 a proactive state model. This should be able to use closed blockchain-based data integrity platform. It will be able to detect and solve all cyber incidents in the public ecosystem within one second. But it will also try to do different things. It will try to pilot traffic accident resolution. Our objective is that after an incident, the resolution will be there in 30 minutes' time, if it is only an insurance case. So if an accident happens and you log it into the system, then in 30 minutes' time all necessary formalities will be handled automatically. For this you would need a fully digitalized insurance system, a fully digitalized car registry, and for people to trust the system, to look for different data and different databases to serve them.

Also, for this we need to look again at our legal space. We need to do something that I do not think any country has even attempted. We need to regulate not technology, but something wider – the human and algorithm interaction, the interaction between a human person and an algorithm. This is what we want to regulate, not any specific technology. Something akin to "you are not allowed to destroy the world", not that you can cannot use more or bigger nukes.

This is what we are discussing, how to do this. We probably need a separate legal entity for AI or AI-similar automated system, how to make a system that to most people looks like AI. I noticed that when I go to forums where AI is discussed, the panel quite quickly boils down to automated systems without actually recognizing, so I think legally at least we do not need to make the distinction.

I think that in two years' time, we have taken another bend around the curve and we will be able to explain to the rest of the world how the world would then look like. I seriously hope that the first bend of safe and personalized internet use will be available for much more people than it is now. It is not only Estonians, it's also Luxembourgish, Finns, Latvians, Danes, many countries who are seeing some success with providing people with digital identity.

Sometimes we are held back by ridiculous problems. For example, I know of a country in Europe, which is really worried that using digital identity and digital space for state -people communication has bankrupted their post office. Which is of course true. Most post offices in the world struggle because no one is sending there emails. I do not know if you have noticed, but most Christmas cards tend to arrive in February because there is a capacity issue at the post offices, they cannot deliver.

I think we really need to be thinking the way we in Estonia are thinking. I am really worried, because previously, when new technology was developed it was normally ordered by NASA or the government or somehow publicly controlled. If you look at digital, it is much more diffuse.

So my bet is that if there is something that fools most of us into thinking that it is AI even if it might not be, it is not going to be government developed. Where it will happen, we do not know. It is quite easy to recognize if someone has really created a singular system, because for that they would need the neuron network-like computer model, these technologies are physical, it is easy to detect, and probably also detect an energy hole somewhere because its energy use will be huge.

I am very worried that our decision-making process at the multilateral level, we are taking much longer to regulate than it is taking technology to develop. Particularly when I look at how the UN has not managed to come to an agreement on how we apply international law in the digital world, despite all the work being done by the working group led by Marina Kaljurand, another Estonian. Or despite the fact that the NATO CCD COE, the Cyber Defence Centre of Excellence has been broadening our understanding of how national law should apply to digital by creating the Tallinn Manual 1.0 and 2.0.

And this is the new situation. With nuclear arms, we knew who had them and we still roughly know who has them, and we have had 70 years to try to come to terms. As I demonstrated, AI could soon solve the problem – but as Trisha here said and discussed before, AI may solve the problem, but will it solve it the way we want it to solve it?