How Estonia transformed into a digital state and is preparing for artificial intelligence. By Kersti Kaljulaid, President of Estonia
Who could have imagined 20 years ago that ‘social media manager’ would be a lucrative job one day? Or a travelling youtuber? Who could have imagined that in principle border guards can monitor their part of the land border even far away from a Mediterranean beach? Technology will keep transforming our lives, destroying and creating jobs and altering the way we work. Efficiency gains from computerization and widespread use of robots could easily supply the material goods we all need with much less human participation in the production chains.
These changes are indeed at least as profound as the industrial revolution, but there is an important difference in the way our society works today. Through education, health care and social systems states can – if they set themselves this ambition – make sure that the externalities of change will not be ignored by society as a whole and be borne only by those vulnerable to change. I believe that the welfare state is up to this challenge, if it can maintain and increase its capacity to provide services in an egalitarian way.
We need to make both our legal framework and people’s skill-set technology-neutral, meaning that neither should be dependent on any concrete contemporary technology as technologies often change and develop quickly. In addition, we must learn to provide our public services as efficiently as Amazon sells books – no physical presence, no cost of application, no opening hours. On top of all that, we must recognize that the new economy will not create taxes the way the old one did and we must radically adapt our tax-thinking.
The Estonian example shows how the public sector can effectively mimic technological change without noteworthy time lag and without falling behind the expectations people have on service quality and accessibility. Estonia is often described as a genuine digital society – for a good reason. Over 95 % of all tax declarations are made online, which means that the state only pays about 0.40 euro to collect 100 euros of tax revenue. The digital authentication and signature system, used by the whole population, has been estimated to save up to two percent of GDP annually.
The underlying driver for this transformation has not been the use of the newest technical solutions, nor any “special” or unique character of the Estonian people – but rather the creation of a permissive legal environment. This – along with the common data exchange layer called X-Road – enabled us to create cross-domain interoperability and to deliver services digitally.
It is true that very often it all boils down to citizens’ trust of e-services, and actually to the trustworthiness of the state itself. Estonia probably had a good position in the 1990s: the re-independent Estonian Republic – unlike the Soviet regime – was something that the people had been long dreaming for and therefore they placed more trust in the state. We have learned that people need to feel safe and secure on the streets as well as in a digital context. They need to have an understanding that the government provides the same level of security in the digital space as it does in whatever kind of physical form of governance.
Countries can increase trust in digital services by raising awareness about the little-understood fact that e-services and databanks are much more secure than paper analogues. The digital format provides much more control over personal data than the paper format, provided that the legal space prescribes clear rules on data assembly, storage and use. Estonian people know that meddling in public databases cannot go unnoticed and the officials know that it is a criminal offence to nose around –and this creates additional trust between citizens, state and e-services.
Similarly, with e-voting, when people can check-in to the system, vote, receive a confirmation e-mail and therefore be sure that their vote was counted, is much safer than voting by post. Yes, there are risks in the digital sphere, but overall it is easier to monitor and control than the risks associated with the paper system.
The same basic principles of creating the necessary permissive legal environment also apply to the question of how to tackle artificial intelligence. A permissive legal environment, which explains clearly the rights and obligations of all parties, facilitates the quick transfer of innovative technologies throughout society, be it for commercial or public use. As we have witnessed, without such legal space, innovation can run into deep trouble, even if it does not break any rule. In Estonia, internet safety has been guaranteed for and by digital signature users since the beginning of the century, and we are looking forward to excelling once again in the AI context.
The Estonian digital infrastructure offers unique opportunities at the state level to use AI in an enabling manner – but only if we find innovative ways to support its development by creating the necessary rules as well. Of course, AI does not exist and may actually never materialize, but if we regulate for AI systems, all automated and autonomous systems will be covered as well. Of course, to be successful through all stages of AI development, we need to regulate in a technology-neutral way, for human-algorithm interactions, and to do this in full compliance of human rights, democratic freedoms and the rule of law, including EU law.
Europe has been a global standards setter in terms of protecting personal data and we should continue being so. The latest example comes from the General Data Protection Regulation, where we have set the global standard for digital privacy. The same should happen with AI – whatever the political order in a particular country, the understanding of reasonable use of AI should stem from the European benchmark. With a cross-sectoral view in mind, we will assess whether there is any specific need to regulate, or whether an added interpretation of existing regulations, for instance, could help to solve the questions raised by new technologies. The approach will be about common dominants rather than regulating AI by sectors. Afterwards, it is up to the people to decide what is appropriate for their country.
Estonia is in the process of drafting a national AI strategy and searching for best-use cases in the public sector. The flexibility of the public sector and regulations is important, as the full potential of the evolving technologies is not yet clear. Our government has formed a special inter-ministerial working group with a more general view to AI and other algorithms of this type.
First, we will take a cross-sectoral approach and come up with a strategy on how to address AI and robotics in the public and private sector across the economy.
Secondly, we aim to raise public awareness of the possibilities of AI, robotics and other new technologies to endorse take-up.
Thirdly, with a cross-sectoral view in mind, we will assess whether there is any specific need to regulate or to adapt existing regulations in order to address the questions raised by new technologies.
The most important component of the puzzle we focus on is human understanding and transparency. Otherwise, the interaction between machines and humanity will play out like in Douglas Adams’s “The Hitchhiker's Guide to the Galaxy” where the “Answer to the Ultimate Question of Life, the Universe, and Everything”, calculated by an enormous supercomputer, was the number 42. Nobody was happy. We have gladly incorporated the digital sphere into our commercial and public lives. Now we are seeking to achieve the same success for society with any existing or future algorithm for the benefit of our democratic, transparent and efficient society