Ott Velsberg, PhD, is the current Government Chief Data Officer, Estonia. The Estonian government famously built e-government infrastructure that essentially created a digital state and digital citizens: the data infrastructure X-Road and a compulsory national digital ID that serve as an interoperability platform for existing decentralized databases and a data exchange layer that can be used by public and private sector actors. The digital ID made it possible for citizens to be identified so that they could digitally sign any contract, access essentially any public service such citizens’ digital medical history, order prescriptions, file taxes, vote, and more. The country was way ahead of its time on digital services. Velsberg joined the government in 2018 and is responsible for what comes next: driving the country’s data and AI policy and initiatives related to the use of data in the public sector. Under his leadership, Estonia has gained a reputation as a trendsetter in data governance, open data, and artificial intelligence. Velsberg, an attendee at the Tallinn Digital Summit which took place Oct. 9 and 10, spoke to The Innovator about the Estonian government’s AI strategy.
Q: When you took on your current role in 2018 the country did not yet have an AI strategy. Please tell us where the country is today on its AI journey.
OV: During the 2018 to 2019 time frame some citizens were critical of AI as they were afraid of losing their jobs, some saw it positively and one-third didn’t really care as long as AI would not negatively impact them. At that time government was critical of AI. Today, according to the latest update of government figures, half of Estonia’s households use AI, a bit over a quarter use it at work regularly, 37% of all public sector employees are using it and in the central government the figure is even higher: over 50%. There is increased uptake because people realize that using the technology increases their productivity. On average a person that adopts some of the services can save up to 4 to 5 hours a week. That is 20 hours saved in a month. By 2030, with the expected increase in automation, the time savings could reach 21 million hours for the government.
Q: How is the government using AI?
OV: We have over 200 use cases. One of my favorites is providing recommendations for people who are unemployed on how to improve their employability and what is the most relevant job for them. For the government itself we are using AI to make the government more efficient, for instance streamlining procurement, drafting regulations or even in areas such as writing briefing notes. For example, an estimated 32,000 hours a year is spent on briefing notes – in two ministries: the Ministry of Justice and Digital Affairs and the Ministry of Foreign Affairs. Some 9,500 memos are written each year. We hope to reduce that time by 80% through the MemoGenAI solution we developed last year. Other concrete solutions include AI-powered satellite image analysis to help us identify what areas have a higher risk of flooding. We are also using AI for customer call analysis to determine why citizens are calling government organizations and identify what are the biggest service delivery challenges. AI gives us insights into what citizens are struggling with and identifies what areas are communications are poor.
Q: What about uptake by the business sector?
OV: If you look at our international ranking, we are the sixth country for AI startups and a strong 5th place in terms of AI use overall. There are a couple of reasons for this. When you look at the private sector it is well above European average on usage of the Cloud. We have lot of SMEs– which includes one person companies – that don’t have other options. Companies that are data and AI intensive have shown on average growth of 16.2% in the last two years, contributing an average €5.7 billion in added value to the economy.
Q: Estonia has announced that it wants to create a testbed for companies to safely test new technologies and business models which are prohibited by current regulations. How will that work and why is the country launching the program?
OV: Currently new technologies can’t be developed or tested properly in many fields, because existing legislation isn’t keeping pace with emerging technologies. Developing novel business models and technologies requires testing in the real world, not just in laboratory conditions. It often conflicts with existing laws, as legislation cannot anticipate technological progress, and is key to boosting an economy’s competitiveness — creating new high-value products, services, and jobs. For example, a robotics company is developing self-service pharmacies that allow for increased efficiency and after-hours access. Current laws do not permit operating a pharmacy without a pharmacist being physically present.
Most existing international regulatory sandboxes have had a narrow focus, mainly providing advice on how to comply with existing laws. Estonia’s Experimentation Framework aims to be Europe’s first cross-sector and broad-based measure for regulatory sandbox development, making the country an attractive destination for international companies to test their solutions or invest in Estonian ventures. The goal is to integrate the Experimentation Framework into Estonian legislation in 2025. After the establishment of sector-specific regulatory sandboxes will begin. This project is led by the Ministry of Economic Affairs and Communications with Accelerate Estonia leading the charge. Accelerate Estonia partners with incubators, accelerators, universities, and organizations to help grow innovative businesses in their portfolios, by guiding them through regulatory hurdles and validating their solutions with top-level leaders in the public sector.
Q: What can you tell us about the Baltic AI Gigafactory and Estonia’s role in it?
OV: One of the things the considerations we have is to build a pan-Baltic and Polish AI GigaFactory – a next-generation compute infrastructure to train and deploy AI models that serve our region’s languages, values, and needs. Research institutions, companies, and investors across Europe are invited to join the initiative as co-developers, strategic advisors, or early-stage funders. Backed by Estonia, Latvia, Lithuania, and Poland, the project targets €3-4 billion total investment, 65% of it from private capital. The European Commission’s ambition is to establish four to five such AI gigafactories across the EU. We want to ensure one of them is here.
Q: What is Estonia doing to ensure that everyone has the necessary skill sets to use and develop AI?
OV: One of my first initiatives was establishing data science in our universities to build competence as we did not have dedicated programs. The objectives of the government is to change the way we educate our population by harnessing what technology allows. We are changing the way we teach and encouraging students to come up with solutions – in collaboration with AI – to enhance their critical thinking. (See The Innovator’s separate story on AI Leap, Estonia’s national program to integrate AI into the country’s public high schools and vocational schools).
We are also doing wide-scale upskilling to help workers understand the value of data driven decision making. We have appointed data stewards inside the government to deal with document and information management, libraries and archives. The employees who went through our training program are now in management positions in ministries as a result. On the AI front we have taught people to do hypothesis testing and be open to changing the scope of the project. We give them access to different open-source tools that they can use, and we encourage everyone to share their results.
Q: What advice do you have for governments wanting to implement AI?
OV: Technology is just one element. Process is important. Start by keeping it small and simple and concentrate on things that will have really concrete practical impact. Think about what you want to achieve. Integration is critical. Avoid stovepipes [systems procured and developed to solve a specific problem, characterized by a limited focus and functionality, and containing data that cannot be easily shared with other systems]. Educate project managers and product owners so they can carry out similar initiatives in the future. Foster cooperation with academia. Ensure the use of AI is ethical with very practical methods. We use a data tracker that looks at how we are processing data and for what reason. Be prepared to give an overview of how you are using the data in various processes. You need to have that really at the core, otherwise there will be a backlash from society.
This article is content that would normally only be available to subscribers. Become a subscriber to see what you have been missing.
