Developments in the field of artificial intelligence in recent years have created significant new opportunities. Many Estonian public sector organisations have started to develop, test and use innovative AI-based solutions. The introduction of such solutions should help organisations to better fulfil their tasks, including provide better quality services, make faster decisions, reduce costs.

A joint audit is on course for completion in early 2026 in cooperation with the twelve supreme audit institutions of EUROSAI, the objective of which is to assess the readiness of the government sector to adopt AI solutions. The National Audit Office participated in the joint audit and prepared an overview to provide a picture of how Estonian public sector organisations are developing and using AI solutions and what the main obstacles are in this area.

According to the list of AI based solutions compiled by the Ministry of Justice and Digital Affairs, 130 solutions have been developed in the public sector, but this list does not give an overview of developments in recent years, and experts believe that many of them are not AI solutions. Approximately 30 AI-based solutions have been created in the organisations that responded to the survey of the National Audit Office. Most of these solutions are still being tested and do not offer significant cost savings, better quality public services or faster decision-making.

The main obstacles to the creation and introduction of AI solutions are: 

• The development of AI solutions is hampered by a lack of awareness among employees of the options offered by AI and its areas of application. On the one hand, the development capacity is limited by the lack of specific technical expertise, for example, the lack of a smart customer from an AI perspective, whose involvement is necessary to develop solutions. Nor do ideas or proposals for new solutions emerge in organisations where the majority of staff have no knowledge of the potential of using AI in their field.
• The creation of AI solutions is hindered by the poor quality of databases. The survey carried out in the course of the overview showed that a considerable number of organisations see data quality as an important issue and are working to improve it. At the same time, the survey revealed that as many as one-third of organisations are not actively engaged in improving data quality, which in turn makes it difficult to develop and implement AI solutions.
• The creation of AI solutions is hindered by the inability to cope with regulatory restrictions. Difficulties mainly arise from data protection rules that limit the use of personalised data both in the training and implementation of AI solutions.

In order to create a better environment for the development of AI solutions, the Ministry of Justice and Digital Affairs, which is leading the area, should pay more attention to removing the obstacles to the development of solutions. At the national level, it is necessary to support public authorities in making the right choices in the legal environment, by developing guidance, carrying out training, giving advice, etc. for this.

Organisations themselves must invest significantly more in improving data quality and increasing knowledge to successfully implement AI solutions. In the future, as solutions are developed and data volumes increase, appropriate IT infrastructure solutions (i.e. hardware and software environments supporting computing and software development) must be found to maximise the performance and security of AI solutions. In addition, organisations must assess whether their planned AI solution is economically reasonable, i.e. whether it will help save costs, improve the quality of public services or enable to make faster decisions.