On 17 October, on-line workshop was organised together by the European Commission's Directorate-General for Communications Networks, Content and Technology and the consulting company ICF. The meeting focused on the development of recommendations for the effective use of artificial intelligence systems in public administration services. The workshop was attended by a representative of the Foundation - Iwona Karkliniewska (AI Researcher).
The main purpose of the meeting was to map the existing problems in the implementation of AI systems in public administration and to develop proposals for solutions and recommendations for decision-makers of legislative work towards adapting the AI ACT and the role and importance of the European Commission in the implementation of legal solutions for artificial intelligence systems.
Workshop was conducted by representatives of the European Commission and OECD as well as ICF employees.
The moderator presented the defined areas of challenges in the implementation of AI systems and asked the participants to map the problems and then propose solutions.
The workshop was divided into two parts.
The first part focused on identifying the threats that have the greatest impact on AI systems in public administration, and then proposing solutions.
The issues were divided into the following problem areas:
1) public procurement;
2) data management;
3) cross-sectoral technological challenges of using AI systems;
4) organizational capabilities of institutions implementing AI systems.
For each of the areas presented above, a question was formulated, and then challenges were defined - problems that need to be addressed in a given area. During the exercise, the participants presented in detail the issues that should be paid attention to in a given area and proposed solutions to the presented areas of issues.
Ad.1 How to meet the challenges in the public procurement process?
The following issues were identified as problematic:
• lack of understanding at the central level about the role of the AI systems used and the areas where these AI management systems can be a good solution; lack of knowledge about the applied AI systems;
• lack of transparent regulation regarding the use of high-risk artificial intelligence systems;
• the lack of distinction between the needs of using AI systems by the public sector and the private sector.
Ad.2 What are the challenges of managing data using AI systems?
Ad.3 How can we solve the challenges related to AI systems in the context of cross-sector activities?
In this area, the organizers defined the following problems:
• limited transparency and measurability of the applied technological solutions;
• efficiency of data management, in particular managing the process of potential bias;
• lack of understanding of the effectiveness of using systems as systems supporting technological development (added value in the process);
• understanding the responsibility and management efficiency of the AI systems used in the area of education, health, and transport.
Ad.4 How should organizational challenges be addressed and capacity-building for institutions implementing AI systems?
- First of all, the attention was drawn to the lack of competence of the staff in terms of knowledge about AI, both in the context of the technical use of the solution and the understanding of the use of the system based on the improvement of management processes in the organization;
- focusing on solutions implementing technological innovations - not taking into account the technological element in the long-term functioning of the organization;
- lack of centralized resources to support the understanding of the process of selecting solutions using AI systems;
- the lack of developed strategic framework - a policy regarding the use of AI systems in the public sector (currently at a very general level).
In the second part of the workshop, participants were divided into groups and, as a result of teamwork, proposed solutions were developed, as well as an indication of the role and importance of the European Commission in the process of activities. The Moje Państwo Foundation had the opportunity to comment on issues related to: regulatory sandboxes, data exchange standards, in particular metadata collections; the effectiveness of the communication tools used and the subject of the so-called project silos.
We drew attention to the need for effective use of communication tools and building a strategy for dialogue between citizens and organizations in the use of AI systems towards citizens. In addition, the issue of the need for a coherent transmission of information on AI, at EU- and national levels, was highlighted.
The solutions developed during the workshop will form the basis for further discussions on adapting the so-called AI ACT.
Iwona Karkliniewska, AI Researcher, Moje Państwo Foundation