Estonia’s National Strategy for Artificial Intelligence
Summary of The Kratt Report and Estonia’s National Artificial Intelligence Strategy for The Period 2019–2021
As described in the subtitle this is a summary of the Kratt Report and Estonia’s National AI strategy for 2019–2021.
The report uses the EU’s definition of artificial intelligence, that is: “…systems that exhibit intelligent behaviour, that analyse their surrounding environments and make autonomous decisions to a certain extent to achieve goals.”
1. The Kratt Report
The report by the expert group was called Report of Estonia’s AI Taskforce. It is a report written by EY and was funded under the Operational Programme for Cohesion Policy Funds, 2014–2020, priority 12 “Administrative capacity”, action 12.2 “Development of quality of policymaking”.
“At the beginning of 2018, an expert group — managed by the Ministry of Economic Affairs and Communications and the Government Office — was established and a cross-sectional coordination project initiated.”
The tasks of the expert group were as follows:
- prepare draft legislation to ensure clarity in the Estonian judicial area and organise the necessary supervision;
- develop the Estonian artificial intelligence action plan;
- Notify the public about the implementation of kratts and introduce the possible options (e.g. the website www.kratid.ee was developed for this, idea gathering was organised, etc.)
Conversations with different parties, discussions in various working groups, background study of foreign countries, and proposals from the companies and experts engaged in the field were used as input for preparing the report
In the report they actually specifically use the word ‘kratts’ throughout the report.
“kratt” in the framework of this project, meaning practical applications based on artificial intelligence technologies (in the narrow meaning of artificial intelligence) performing a specific function
For Estonia to stand out and to achieve a competitive edge in the AI race, the team of experts created the so-called “Kratt report” — named after a mythical Estonian creature that is devoted to serving its master, but can become bad if left idle…
The Estonian definition of “Kratt” stands for a practical application that uses artificial intelligence and that fulfils a specific function.
In Estonian mythology, a Kratt is a magical creature. Essentially, Kratt was a servant built from hay or old household items. Therefore, the Estonian government uses this character as a metaphor for AI and its complexities.
While classical software solutions are programs that fill orders given to them by the programmer, AI algorithms don’t have a predefined program logic and “Kratt” must come to the right conclusion through machine learning.
“Kratts = AI in Estonia”
The most fun wordplay that I found in the early suggestions were the following:
“Bureau-Kratt” — concept of a personal virtual assistant through collaboration across networks.
It sure made me chuckle, smile and face-palm.
It does show the Estonians in the administration or EY have some humour!
The report concludes that Estonia should start piloting ‘kratts projects’.
According to the report Estonia needs to:
- Act as a pioneer in piloting kratts solutions
- Awareness needs to be raised to foster implementation of kratts in the business sector
- In order to introduce kratts on a wider scale, funding of R&D needs to be increased (the goal is 1% of GDP), a competitive grant for doctoral candidates needs to be established, and investments into IT need to be increased significantly.
- Not make any substantial changes in the basics of the legal system and there is no need for a so-called harmonised “kratt act”. Only slight changes in liability.
“On the one hand, this expert group report has been created to boost the implementation of kratts in the short term (until 2021). On the other hand, the expert group sees the need to establish a long-term strategy for artificial intelligence in Estonia that would take into account the experiences gained from the initiatives prepared under the current action plan.”
They are establishing as well a working group to monitor the effects of the short-term implementation.
Specific Suggestions To Sectors From the Kratt Report
The report had specific suggestions for each sector.
For the public sector:
- Order basic components and “tools” for kratts and make them available in a reusable format.
- Establish a cooperation network of data science and kratts in the public sector.
- Establish and disseminate instruction materials for initiating and carrying out kratt projects (including for responsible development and sustainable operation).
- Organise the dissemination of knowledge and exchange of experiences by introducing the possibilities of kratt projects and conducted projects in various networks and formats.
- From the aspect of the sustainability of the Estonian language and culture, it is important to strengthen the implementation of language technologies. l At least at the level of ministries and areas of government, establish and fill the positions of chief data officers.
- Ensure flexible and sufficient funding means for test projects, including kratt projects, in the funding measures for the e-state development.
- Develop technological sandboxes for the testing and development of kratt applications in the public sector.
- Prescribe technical requirements for ensuring sustainability in the conditions for funding developments of kratt projects.
- Launch deep learning workshops for data management and create support opportunities for conducting data audits in institutions
- Create and test the interoperability of kratts and the concept of a personal virtual assistant (i.e. #bürokratt in Estonian).
For the private sector:
- Determine an organisation developing the field of artificial intelligence to increase awareness thereof.
- Create a training programme for the implementation of machine learning.
- Order a web-based training programme (MOOC) to increase awareness of kratts on a wider scale.
- Extend industry digitalisation measures to other sectors in addition to the manufacturing industry in a scope of up to 50 million euros.
- Establish an innovation grant with a budget of up to 100,000 euros per project to create a proof of concept for solutions developed based on state databases.
- Support the implementation of kratt-based pilot projects with up to 50,000 euros and 30% own contribution.
- Support the implementation of kratt-based products with up to 200,000 euros and 30% own contribution
For the R&D + education:
- Make significantly higher investments in higher education in the field of ICT and research and development activities. This includes, among others, increasing the number of ICT lecturers and expanding artificial intelligence studies at master’s and doctorate level, as well as providing it horizontally in other fields.
- Establish an artificial intelligence-oriented DIH in Estonia and support its launch from the state budget.
- In addition, Estonia should be actively joining the EuroHPC² projects which enable institutions and businesses engaged in research and development access to high-performance computing capabilities
Estonia has grown fast, but growth has not reached the expectations and potential.
Solutions Already Introduced in Estonia
Examples of solutions that have been introduced:
- Unemployment Insurance Fund — profiling jobseekers and generating suggestions for open positions;
- Police and Border Guard Board — prediction of locations where highway patrols are required;
- City of Tallinn — detection of the volume of traffic, using cameras.
- Estonian Agricultural Registers and Information Board (ARIB) — detecting whether meadows are being mowed to monitor the use of supports;
- Estonian Road Administration — prediction of the condition of roads and, therefore, return on investment;
- Health insurance fund — prediction model for the health of chronically ill patients and their treatment requirements;
- Statistics Estonia — chatbot in customer service;
- Information System Authority (RIA) — detection of security incidents by monitoring traffic in the X-Road;
They recommend that a:
“…wide-scale approach that is already under development should be fostered and as many projects in as many areas as possible should be supported.”
They mentions a set of different areas with existing and additional measures in public sector.
- Raising the awareness of the management and officials
- Improving the skills of officials
- Ensuring funding for the development of kratts
- Simplifying the technical launch of projects
- Ensuring the sustainability of kratt solutions
- Increasing the availability of data that’s suitable for development (including open data)
- The concept of #bürokratt: ensuring the interoperability of kratt solutions and creating a virtual assistant for the e-state
In addition to this they made a list of developers and implementers of artificial intelligence solutions in Estonia.
- Milrem Robotics Driverless crawlers that can drive on any landscape
- Mindtitan Advanced solutions using artificial intelligence components
- LeanEst Comprehensive solutions where image recognition is based on artificial intelligence. For example, this is used in the timber industry to detect and categorise defective planks.
- Alphablues Chatbot
- LHV Chatbot
- Elisa Chatbot
- Proekspert Industrial automation, smart environmental solutions, and data science
- Mindtitan Consultation undertaking specialised in artificial intelligence solutions
- Sifr Consultation undertaking specialised in artificial intelligence solutions
- Smart Load Solutions Building automation for underfloor heating solutions
- Taxify (Bolt) Ride sharing service
- Fujitsu Predictive device maintenance in the industry, health technologies
- FlowIT Solutions for automating activities, algorithms predicting purchase orders
- Datel Data interpretation and visualisation solutions for the state geographical information system application. For example, supplying the Estonian Rescue Board with geographic data, landscape analysis
- Cleveron Last mile logistics and solutions for parcel machines
- Skillific Automatic matching of profiles of jobseekers
As with public sector there is a list for improvements in private sector:
- Raising awareness on kratts
- Support of investments made by companies to apply krattbased products
- Support of investments made by companies to develop kratt-based products
As well as for R&D:
- Expanding the education of specialists in the field of artificial intelligence
- Research institutions and the capacity of R&D
Towards the end of the report there is mostly repetition of other legal frameworks presented by the European Union.
The state is liable for damage caused in private relationships pursuant to the principles of civil law. Creating a kratt that poses a threat to property and/or the environment should be regulated as a punishable act.
2. Estonia’s National Artificial Intelligence Strategy for The Period 2019–2021
The current AI Strategy in Estonia is largely based on the Kratt report with some added action points.
- Action item
- Responsible agency
As such it is one of the most specific strategies I have seen so far.
According to the current strategy and based on existing knowledge, Estonian government will invest at least 10M euros in 2019–2021 to implementation of AI strategy in its different directions.
“This current strategy has been prepared based on these proposals, as a plan on how to implement the expert group’s recommendations. It was adopted at Cabinet meeting on 25 July 2019. The strategy is a sum of actions that Estonian government will take to advance the takeup of AI in both private and public sector, to increase the relevant skills and research and development (R&D) base as well as to develop the legal environment. This strategy is also Estonian national AI strategy in the European Union’s coordinated AI action plan context, synchronised with and supporting relevant EU-level activities.”
The categories do to some degree match the ones in the Kratt Report:
- ADVANCING THE UPTAKE OF AI IN PUBLIC SECTOR IN ESTONIA
- ADVANCING THE UPTAKE OF AI IN PRIVATE SECTOR IN ESTONIA
- DEVELOPING AI R&D AND EDUCATION IN ESTONIA
- DEVELOPING LEGAL ENVIRONMENT FOR UPTAKE OF AI
I will go through all the actions and attempt to condense the projects at a later time.
This is #500daysofAI and you are reading article 211. I am writing one new article about or related to artificial intelligence every day for 500 days. My current focus for 100 days 200–300 is national and international strategies for artificial intelligence.