Research

In recent years, the Estonian government has taken several steps to support the implementation of artificial intelligence (AI) in both public and private sector institutions. In Estonia, the term “Kratt” is used in relation to AI, and it refers to practical applications that use AI to perform specific functions. Automation related to AI is a major focus of discussion, both in the government and in the public domain.

While there are discussions about the use of automated decision-making or AI, for the moment, the main focus is predominantly on the benefits these technologies will bring. In the public sector, the benefits of AI are seen as threefold: it increases the user-centeredness of services, improves data analysis, and makes the country work more efficiently to achieve the goal of developing an e-government. AI can also play an important role in the digital revolution of industry and to attract new investment and innovation in Estonia – technology developers are searching for a development and test environment that favors AI solutions. In this regard, the Estonian strategy seems to focus on a small, well-established, digital e-government, and on a future that sees the country become a good testing ground for AI solutions.

Several research centers in Estonia focus mainly on the technical aspects of AI or offer a testbed to pilot the solutions to possible problems. In addition, a study that will focus on the opportunities for using AI and ADM in e-governance is also planned. There are several governmental and non-governmental organizations which have pointed out possible risks and problems associated with automated decision-making. Since 2017, Estonian residents can check the Eesti.ee website to see which institutions have accessed their data and for what purposes by using the Data tracker tool. By September 2019, four major government agencies were participating in this project (The Population Register, The Health Insurance Fund, the Estonian Unemployment Insurance Fund, The Social Insurance Board). The website also indicates if data has been accessed for automatic processes or by a specific automated service. This solution is seen as vital in fostering trust and transparency in governmental services (Plantera 2019).

Since 2018, municipal governments in Estonia have been able to use a tool called The Youth Guarantee Support System (YGSS) (Mis on noortegarantii tugisüsteem? – Tööelu.ee o. J.). This tool helps case managers (social workers, child protection officers, youth workers, etc.) identify young people aged 16-26 who are not in education, employment or training (NEET). It is used to support them when they need to return to education or enter the labor market. Case managers of the Youth Guarantee Support System or the municipal government employees can see information about young people living in their specific locality. However, up until now, not all of the municipalities have joined the program. In order to join, the municipal governments have to send an application to the Municipal Government Information System for Social Services and Benefits (STAR) (Noortegarantii | Sotsiaalministeerium o. J). The case managers or municipal government employees are then shown a list in the STAR system of all the NEET youth who are registered as residents in a specific municipal area and who – according to other registry data – need help. This information is based on data from nine different registries, creating, therefore, a very comprehensive dataset that includes information on the young person’s address, email, phone number, educational background, and if they studied in Estonian or Russian. Young people have the right to forbid the analysis by a case manager, and in cases where the data is processed, young people must give their consent.

The case manager then contacts the young people identified by the system and a letter or an SMS is sent to each person asking them to complete a survey. The Youth Guarantee Support System Request is automatically triggered twice a year (15.03 and 15.10).

In 2015, the Estonian Health Insurance Fund (EHIF) started cooperating with the World Bank to develop and pilot a risk-based management model for healthcare (Enhanced Care Management: Improving Health for High Need, High-Risk Patients in Estonia 2017) that would help increase the integration of health services. Risk-based treatment management enables family physicians to identify patients on their list with multiple chronic illnesses who would benefit most from additional help with prevention, counseling, and follow-up care to improve their quality of life. If these patients are neglected by family physicians, this can lead to serious problems, including unnecessary deterioration of health, which, in addition, causes unnecessary costs for the healthcare system (avoidable hospitalization, duplication of studies, etc.). The aim of the machine learning project was to improve the so-called mechanically constructed algorithm developed during the Risk Patient Pilot Project and to better identify the patients who should be included in the risk patient program.

Between 2016-2017, the Estonian Health Insurance Fund (EHIF) and the World Bank conducted a pilot project on treatment management in Estonia. The first step for this was to develop a clinical algorithm in cooperation with family physicians in order to identify patients who patients with selected diagnoses are likely to be admitted to the hospital. The solution first identifies certain medical conditions/diagnoses in the EHIF’s medical invoice database. It then provides a practical model for family physicians to predict which patients are more likely to be admitted to the hospital or suffer other health problems. The EHIF evaluation report from 2017 (p.33) also emphasizes that: “the prioritisation of patients within the registry list based on behavioral data (i.e. whether the patients have filled all their prescriptions during past months) and social patient characteristics (e.g. whether the patient may be socially vulnerable) still needs to be fully developed and then used, as currently this information is not used in the patient selection process”.

The EHIF has been cooperating with the World Bank since 2014. The pilot was created by the World Bank with the involvement of the Estonian Health Insurance Fund. Costa Rica is also involved in the project by allowing access to its medical billing database (to look at clinical/socio-economic data not included in EHIF databases).

The pilot was officially launched in January 2017 (Enhanced Care Management: Improving Health for High Need, High-Risk Patients in Estonia 2017). Throughout the pilot, family practitioners joined a series of webinars, led by the local pilot coordinator (World Bank consultant), to reinforce and refresh the initial training. This solution contributes to the empowerment of primary care and also helps find the best algorithm to predict which patients with selected diagnoses are likely to be admitted to the hospital. The solution first identifies certain medical conditions/diagnoses in the EHIF’s medical invoice data- base. It then provides a practical model for family physicians to predict which patients are more likely to be admitted to the hospital or suffer other health problems. The EHIF evaluation report from 2017 (p.33) also emphasizes that: “the prioritisation of patients within the registry list based on behavioural data (i.e. whether the patients have filled all their prescriptions during past months) and social patient characteristics (e.g. whether the patient may be socially vulnerable) still needs to be fully developed and then used, as currently this information is not used in the patient selection process”.

The EHIF has been cooperating with the World Bank since 2014. The pilot was created by the World Bank with the involvement of the Estonian Health Insurance Fund. Costa Rica is also involved in the project by allowing access to its medical billing database (to look at clinical/socio-economic data not included in EHIF databases).

Estonian start-up Skillific is developing a recruitment service (Skillific kasutuslugu | Krattide veebileht o. J.) that uses a machine learning software program to enable employers to find potential employees by analyzing the skills and knowledge they need. Skillific’s search process is based on, among other things, the European Skills, Competences, Qualifications and Occupations (ESCO) classification system, developed by the European Commission, which defines skills needed in many areas of life. Based on this, the machine learning algorithm looks for potential candidates in the same skill category as the job profile. Skillific’s AI application, through the web, partner data, and public databases, can find and profile potential candidates for different jobs, and also assess their suitability for those jobs (Eestis töötati välja tehisintellekti sugemetega talendiotsingumootor Skillific 2016).

At the moment, the software is still in the testing phase as there is not enough good quality training data for the full implementation of the software. The Skillific database already has about 400,000 user-profiles and a smaller number of workplace profiles. Based on this information, the machine learning algorithm predicts the suitability of job candidates and provides better quality solutions through easier storage and collection of data. Without this system, the process is very resource-intense and generally does not utilize previously analyzed candidate profiles.

The aim in the future is to create an environment similar to the Finnish database, MyData. With the consent of the job seeker, this system would allow companies to access more data and thus provide a better service to both the job seeker and the employer. Skillific estimates that the entire recruitment process could be fully automated within the next 5 to 10 years.

Estonia’s Unemployment Insurance Fund (EUIF) increasingly uses different automated solutions in its everyday work (Kõik muutub automaatseks: töötukassa IT-osakond ei istu päevad otsa mustades ülikondades, tõstes tuimalt pabereid ühest hunnikust teise 2019). After a citizen registers on the Unemployment Insurance Fund website as unemployed, the data is checked and, if it is correct, the citizen is then registered as unemployed. In the background, the system uses AI to check an applicant’s data in different databases. It then decides which document to send to the applicant.

EUIF also uses ADM to decide what a person is entitled to, including what amount of unemployment aid or unemployment insurance aid, and for how long (Raudla 2020). According to Erik Aas, an EUIF council member, 50% of those decisions are made entirely through ADM (Raudla 2020).

In addition, the Unemployment Insurance Fund plans to start predicting how long specific individuals will remain unemployed.

The main goal of this is to make use of all of the data the Unemployment Insurance Fund has.

According to the Estonian Unemployment Insurance Fund development plan for 2019-2022 – which aims to increase the impact, quality and accessibility of services – a profiling model should be used. This would help to identify the risk of long-term unemployment for each client and to choose the appropriate counseling channel, the frequency, and specific services. The Estonian Kratt report (p.18) indicates that this profiling solution is already in use.

The Unemployment Insurance Fund has the right to use automated decision-making (Millal teeb töötukassa Teie suhtes automaatse otsuse | Töötukassa o. J.) according to the Unemployment Insurance Act § 23 lg 4. The applicant is informed of this ADM process when submitting an application and the use of ADM is also displayed on the decision document.

The machine vision Kratt, or AI solution of the Public Transport and Traffic Management Department in Tallinn, started in autumn 2018 in cooperation with Sifr OÜ (Kasu- tuslood | Krattide veebileht o. J.). The main purpose of this solution is to monitor the traffic load In Tallinn, specifically the cars driving in and out of the city every day.

This information is used to make decisions about parking problems or road construction, among other issues. To test the solution, three cameras were selected to monitor intersections in Tallinn. This helped train the software to count vehicles passing the camera. This solution uses a machine vision algorithm, which counts buses, cars, trucks, and motorcycles. In the future, it is hoped that the same solution can be used to count pedestrians. The accuracy of the results can be affected by different weather conditions, including fog, and by dirt on the cameras.

 

/ Age-verification technology used in Kelia

In Keila, a city in northern Estonia, the Keila Consumer Co-operative uses a self-service check-in desk that verifies the customer’s age with special age-verification technology and is designed so that the customer can buy age-restricted products. The self-checkout ID is based on Strongpoint self-checkout software and the Yot Digital Identification Platform (Keila Tarbijate Ühistu võttis kasutusele Euroopas ainulaadse näotuvastusega iseteeninduskassa – StrongPoint J.). To purchase age-restricted products, the customer’s age is checked at the self-checkout desk and after that the customer can then pay for the item. If the customer wants to buy tobacco products, they have to choose the item from the self-service checkout and then pass a preliminary face detection process which issues a paper to allow the customer to go to the tobacco machine. The vending machine then performs another check before issuing the tobacco. The customer does not have to provide any additional documents to verify their age.

This is believed to be the first solution of its kind in use in Europe.

A journalist from the digital newspaper Delfi conducted an experiment (Veltman 2019) to see if the self-service check-in desk face recognition software can distinguish adults from minors. For this purpose, he recruited 14 and 15-year- old girls and boys to try to buy age-restricted products. Initially, it seemed that the technology stopped minors from purchasing age-restricted products, correctly deducing that they were underage. However, on the second attempt, the technology permitted an underage boy to buy the products. The software developer explained that the anomaly happened due to the repeated face detections without completing the purchase the first time. According to them, the problem should be fixed now.

However, several big supermarkets are not ready to start using this software as, for them, this system is still in the development phase and may be a security risk (Belkin 2019).

Education services in the city of Tartu use a system called ARNO (www.arno.ee) for submitting and processing applications for places at nursery schools and for childcare. Tartu uses ADM systems to determine a child’s municipal school based upon the child’s address registered in the population register.The school placement offering is generated automatically once a year and is based upon three sets of data: the automatic queue for a child of pre-school age, the parent’s preferred kindergarten and the nearest kindergarten. Parents can apply at a later stage to change school depending on specific circumstances. In the case of kindergartens, parents have to apply for a kindergarten spot after a child is born. ARNO is developed by OÜ Piksel.

In 2019, the Social Insurance Board (SKA) completed its first automatic proactive service (Sotsiaalkindlustusamet avas esimese proaktiivse avaliku teenuse | Sotsiaalkind- lustusamet 2019) meaning that parents no longer need to apply for services. The service is based on the principle that, as the state already has the information about each citizen from birth, all the following services, such as family benefits, can be activated automatically. It is the first event-based service in Estonia, and the SKA aims to move fully towards application-free services.

Once a birth is registered in the population register and the child is given a name, the Social Security Agency will send an e-mail to the parents. Once the parents have confirmed receipt of the notification, they can receive family benefits.

This solution was developed by Nortal and is considered to be unique in the world. The service is based on an automated algorithmic solution, and it has also led to changes being made in some organizational processes (Sotsiaal-kindlustusamet avas esimese proaktiivse avaliku teenuse | Sotsiaalkindlustusamet 2019).

In 2019, the Ministry of Justice officially requested that Estonia's chief data officer design a robot judge to preside over small claims disputes of less than 7,000 euros to clear a backlog of such cases (Niiler 2019). The project is still in progress, but in theory, the two parties in each case will upload all the relevant information to a database that the AI software will then analyze to render a decision based on pre-programmed algorithms and previous training. The software’s decision would be legally binding, but that decision could then be appealed and looked into by a human judge. The country’s economic ministry is considering granting AI and robot judges legal status to help the legislation allocate responsibility for decision-making involving AI-controlled software. This solution is still considered an aid to the courts in Estonia and not a standalone solution.

Still, there have been some changes made to Estonian procedural law which will be introduced in several stages between the years 2020-2023 (Mandri 2019).

The AI Programme, called the “Kratt plan” (Majandusja Kommunikatsiooni ministeerium/Riigikantselei 2019b), was commissioned by the Government Office and launched at the beginning of 2018. A steering group consisting of experts from public and private sector organizations and academia was established and is led by Siim Sikkut. Kratt is a magical creature from Estonian mythology and it was decided to use this character as a metaphor to facilitate communication. In Estonian mythology, Kratt was a servant who built things from hay or old household items and he needed to be looked after so that it would not go idle. AI is seen as a huge opportunity and a powerful tool for Estonia which also needs to be wielded responsibly and safely so that the Kratt project, like the mythological character, will not come to harm if left unattended.

The main tasks of the expert group were to: prepare draft legislation to ensure clarity in the judicial sector, to organize the necessary supervision, to develop an AI action plan, to notify the public about the implementation of the Kratt project, and publicize possible options (e.g. via development of the www.kratid.ee website).

The proposals made by the expert group were compiled into a report (Majandusja Kommunikatsiooni ministeerium/Riigikantselei 2019b) that outlines advice and activi- ties on how to accelerate the use of AI in both private and public sector organizations. The proposals made by the expert group in 2019 formed the national AI action plan for 2019-2021 (ajandus-ja Kommunikatsiooni ministeerium/ Riigikantselei 2019). The strategy was adopted at a cabinet meeting on 25 July, 2019. According to the strategy, the government takes a leading role in accelerating and supporting the use of AI-based applications in both the public and private sectors alike. The Estonian government will be investing at least 10 million euros between 2019 and 2021 to implement the strategy.

The proposals, and Estonia’s AI action plan, focus on developing the basic competencies required for implementation. Moreover, it is emphasized in the planning document and on the e-Estonia webpage, that “Estonia could become the role model and testbed for the rest of the world as a place where Kratt, or AI, is put to work for the people’s well-being in both the public and private sectors”.

In particular, Estonia emphasizes (Majandusja Kommunikatsiooni ministeerium/Riigikantselei 2019b) the objective in the AI plan should be to boost AI development in the country, starting with basic competences. The plan gives an overview of the development of AI and of the plans in other countries and, based upon this, it recommends that particular effort should be made to facilitate the implementation of Kratts in the public sector.

The 2019-2021 action plan (Majandusja kommunikatsiooniministeerium/Riigikantselei 2019a) not only considers the steps that should be taken to boost AI development in Estonia, but also looks at support for research, improvement of skills for public sector workers and financial support. One of the actions includes making basic Kratt components and “tools” available for reuse. This involves establishing a collaborative network in the public sector, and creating and disseminating instructional material for launching and conducting research projects. It also emphasizes the need to showcase the opportunities AI brings and introduce specific projects in different networks and formats. Furthermore, it suggests that, at least at the ministerial level, roles for chief data officers should be established. It also emphasizes the need to launch in-depth data management workshops and create support for data audits in institutions.

The legal analysis in the report concludes that there is no need for substantial changes to the basics of the legal system and no need for a specific “Kratt act”. However, the report emphasizes that current legislation needs several changes. For example, there needs to be clarity around the issue of liability related to the implementation of Kratt. There is legislation to regulate automated decision-making and profiling. A specific institution can only use ADM or profiling by using a specific legal act, a signed contract, or if a person gives consent.

The report concludes (Majandusja Kommunikatsiooni ministeerium/Riigikantselei 2019b) that Estonia should start piloting Kratt projects to gather initial feedback and experience to better inform long-term planning. This is seen as an important step as pilot projects will help the government better understand the possible benefits and risks of using Kratt while also helping to form future strategies. The strategy also emphasizes the need to raise awareness, increase research and development activities, encourage innovation, and to foster the implementation of AI, or Kratt, in the business sector.

Although the Kratt project specifically does not mention ADM, many of the activities financed and supported through the AI strategy for 2019-2021, are also important in relation to ADM.

Moreover, there is a plan to create a bureaucratic Kratt, or artificial solution, as a personal digital assistant. The idea behind this AI solution coincides largely with the development of the Aurora concept in Finland. It would, therefore, be reasonable to cooperate between these developments – or at the very least, exchange ideas and experiences and to develop technical solutions together, if possible.

Among other things, this would enable interoperability of the state Kratt ecosystems of Estonia and Finland so that the virtual assistants, or Kratt sessions, could also function as cross-border services. The common digital space shared between Estonia and Finland is one of the priorities of the digital policy of Estonia and this could also continue in the Kratt age. However, this is still in the early stages of development.

As of October 2019, there were at least 23 AI solutions deployed in the Estonian public sector and that number was set to increase to at least 50 by 2020. The examples already in use show that the scope of the use of Kratt, or AI, has some limits. Kratt solutions are not used for making decisions on their own yet, but they are used to help humans make faster and better decisions.

Estonia’s AI strategy emphasizes the need for a broader understanding of AI. Therefore, one of the aims the expert group posed in the AI strategy document was to develop a web-based training program. The Finnish “Elements of Artificial Intelligence” online course, which was developed by the Department of Computer Science at the University of Helsinki in partnership with the technology company Reaktor, was used as the basis for a suitable online course. In the middle of November 2019, an Estonian language version of the “Elements of AI” (Tasuta veebipõhine sisseju-hatus tehisintellekti mitte-ekspertidele o. J.) online course was launched. The Estonian version of the course was developed by Tallinn University of Technology (TalTech). According to the strategy, the main aim of the course is to increase the interest and receptivity of AI in Estonian businesses. On the course webpage, it is also emphasized that taking part in the course will help citizens see AI as a friend, something that will offer support in life and not something to be feared.

The course introduces basic concepts and applications of AI and machine learning to increase knowledge and equip people with a good understanding so that they can participate in the debate on the subject.

The societal implications of AI, such as algorithmic bias and de-anonymization, are also introduced in the course. This helps underscore the need for policies and regulations to guarantee that society can adapt well to the changes the increasing use of AI brings. The course is free for anyone to take and close to 220,000 people have already completed it.

The aim (Liive 2019) is to spread knowledge about AI to at least 1% of Estonian citizens.

The former president of Estonia, Toomas-Hendrik Ilves, is the course patron. The course has more than 30 partners from different businesses and organizations.

To promote the course among Estonians, Tallinn University of Technology started a campaign – #AIväljakutse (#AIchallenge). More than 1200 people took the challenge when the course was first launched. The challenge to complete the course has been taken up by nearly 35 companies and organizations in Estonia who want to educate their employees about AI.

The Center of IT Impact Studies (CITIS) was established in 2015 and it is an integrated teaching and research center at the Johan Skytte Institute of Political Studies at the University of Tartu, Estonia. Their main aim is to use big data generated by various Estonian public services (for example, e-health, digital ID, e-residency, etc.) to estimate the impact those services have economically, politically, and socially (Center of IT Impact Studies (CITIS), n.d.). CITIS teams also contribute to the development of the e-state directly by creating and testing prototypes for new e-services, based on the cross-usage of existing data. The CITIS team, together with Nortal, one of Estonia‘s largest developers, are building a set of profiling and segmentation policy tools for the Estonian Unemployment Insurance Fund (EUIF) (Profiling and policy evaluation tools for EUIF (2018-2020) 2019).

Established in 2009, STACC is the leading machine learning and data science competence center in Estonia and it also develops artificial intelligence solutions. Its mission is to conduct high-level applied research in the field of data science and machine learning in cooperation with a consortium of scientific and industrial partners. STACC was founded by two universities – the University of Tartu and Tallinn University of Technology and by several leading IT companies: Cybernetica AS, Reach-U AS, Nortal AS, CGI Eesti AS, and Quretec OÜ. One purpose established in the Es tonian national AI action program is to direct and support STACC activities related to AI. STACC is one of the six technology competence centers currently in operation and it is financed by Enterprise Estonia. The centers aim to motivate companies to develop innovative products through cooperation with research institutions, thereby bridging the gap between scientific and economic innovation.

The Government Technology and Artificial Intelligence Lab (GovAILab) was opened on 18 November 2019 at Tallinn University of Technology (TalTech). Its main focus is to advise and help governmental institutions find new technological solutions, but it also functions as a place to pilot public sector ADM or AI projects to help decide how to proceed. The lab is seen as a partner for the organization’s leaders, and it helps specialists, who are accountable for the development of technical AI solutions, to make better choices and to experiment with technology.

The Foundation for Future Technologies (FFT) (Tulevik 2018) is a non-governmental organization that aims to help maximize the positive impact of future technologies on Estonian society and to make their deployment a considered and widely understood decision (Tulevik 2018). The Foundation for Future Technologies member Kaspar Kruup spoke at the TEDxTTÜ event on 19 March 2016 about AI. The foundation has also organized panels on topics such as, the future of work, virtual reality, and AI at the 2016 Opinion Festival. In addition, foundation members also organized an AI Think Club in which 75 people took part and they maintain a Facebook Group on Artificial Intelligence and SINA (Tehisintellekt ja SINA FB page o. J.). The Facebook group points to some foreign news articles and discussions but there are no Estonia specific discussions. The page is followed by 375 people. The foundation no longer takes an active part in AI or ADM related discussions.

Two members of the foundation (Allan Aksiim and Pirko Konsa) led the expert group on self-driving cars (Isejuhtiva sõidukite ajastu algus – ekspertrühma lõppraport 2018) which was initiated by the Estonian Ministry of Economy and Communication and the Government Office. However, the work of this expert group has now ended.

The Estonian Human Rights Center is an independent non-governmental organization that stands for human rights in Estonia. In 2019, the center published a document predominantly aimed at mapping the human rights situation in relation to the information society and technology (Käsper/ Rajavee 2019). Authors carried out focus group interviews

and discussions with experts, from both private and public sector organizations, third sector activists, and a journalist, who had come into contact with the topics covered in the report. Several other documents were also analyzed for this purpose. The report considers issues such as profiling by the state, GDPR Implementation, and freedom of expression on the Internet. The report indicates several risks and interviewed experts to determine the possibility of discrimination through profiling and privacy concerns and whether or not proactive services should be considered. The authors (Käsper/Rajavee 2019) suggested developing visual solutions that would make it easier for people to understand which state services use their data and how they use it, but they also emphasized the need for further analysis about how ADM impacts human rights.

The Estonian Institute of Human Rights was established in 1992 and it aims to collect, systematize, analyze, and disseminate information on individual rights for public awareness. On 10 December 2019, the institute organized a human rights conference in Estonia on the topic of “Values and Interests In parallel worlds” (Inimõiguste aastakonverents 2019. 2019). One of the panels discussed how technological developments override human rights and the possible impacts and risks associated with ADM, among other things.

In 2019, the Estonian Research Council signed a contract with a research consortium to look for opportunities where and how ADM could be used in developing public services (Teadlased asuvad uurima tehisintellekti kasutamise võimalusi avalike teenuste pakkumisel – Sihtasutus Eesti Teadusagentuur 2019). This study is funded by the RITA program “Opportunities for the Use of Artificial Intelligence in E-Governance”, jointly conducted by the University of Tartu, Tallinn University of Technology, Cybernetica AS, and STACC and will last for two years. The research results will be used by the Ministry of Economic Affairs and Communications and the State Chancellery for the development of e-government.

The overall aim of RITA is to increase the role of the state in the strategic direction of science and the capacity of research and development institutions to carry out socially relevant research.

The Estonian public sector is smaller than average, and the services provided by the state and local governments must be equivalent. At the moment, the provision of services in Estonia is not optimally organized. One way to improve the quality of services without increasing the number of officials is to implement artificial intelligence, or ‘chats’. The results of the survey are expected by the end of 2021. The budget for the study is 805,212 euros, which comes from the European Regional Development Fund and the state budget of Estonia through the RITA program.

In Estonia, several solutions that use ADM have already been developed and are planned under the Kratt project. This report, and the different organizations mentioned above, emphasize several issues, such as possible discrimination related to data or ethics, related to AI and ADM which should be taken into account. The overall public discussion tends to concentrate on the possible opportunities that developed solutions and AI will bring. Risks and problems are acknowledged, but they are not discussed widely.

Several of the services could be described as proactive as their main aim is to monitor or select citizens who may require additional counseling or other kinds of help. There is one particular proactive service already offered for families, and several others are under development. Services like these can be seen as positive as citizens are provided with the services, they are entitled to without the need to apply for them. Therefore, some of the groups or individuals who would have a right to those services, but who may not be digitally literate, are still offered the services proactively.

In some cases, it can be incomprehensible for a person to understand how the systems reach specific decisions. Although, for some time now, Estonians have had an opportunity to check on the eesti.ee webpage to see which institutions have accessed their data and for what reasons. However, as the Human Rights Center report emphasizes, clearer, visual solutions should be developed. The solution used right now does not offer a clear understanding of what profiling is done by the state, which data is collected, how it is used, and for what purpose. Moreover, it is not always clear if data is used as a part of an automatic process or viewed by an official.