This report addresses the following objective of AI Watch: it presents a summary of the proceedings of the workshop of the 4th AI Watch Peer Learning Workshop on AI use & impact in public services. The event took place online on 28 October 2021, with the participation of 86 representatives from 24 Member States and Associated Countries, amongst which 55 public servants from Member States and Associated Countries, including 14 EU officials, and 17 researchers from all over Europe.
The Workshop was an important opportunity to
- i) engage with relevant stakeholders to better understand the potential use and impact of AI for the public sector;
- i) present and discuss a "Road to the adoption of AI by the Public Sector”;
- iii) validate a set of recommendations with a number of actions addressing stakeholders at different operational levels, aiming at fostering the adoption and use of AI by the public sector throughout Europe
The Workshop was divided into three main sessions.
In the introductory session, the background, the elaboration process, and the content of the draft document of the "Road to the adoption of AI by the Public Sector" was presented. The main content of the document presented at the event consists of a set of 16 recommendations to foster the adoption and use of AI by the European public sector.
The resulting recommendations are based on the analysis of existing research literature, an updated analysis of Member States and Associated Countries national AI strategies, and the results of an EU-wide survey addressed to practitioners from National Public Organisations throughout Europe, and EU public officials dealing with AI adoption in the public sector.
The recommendations have been divided into four Intervention Areas clustering a number of Recommendations, and possible actions addressing stakeholders at different operational levels:
- Area 1. Promote an EU-value oriented, inclusive and human-centric AI in the Public Sector;
- Area 2. Enhance coordinated governance, convergence of regulations and capacity building;
- Area 3. Build a shared and interactive AI digital ecosystem;
- Area 4. Applying value oriented AI impact assessment frameworks.
In the second session, all participants were divided into four groups each assigned to one of the above Intervention Area. Working in breakout sessions, the groups elaborated feedback on strengths, weaknesses, and suggestions for improvement for each of the recommendations in the corresponding Intervention Area. The feedback was then presented by each group in a plenary.
In the third session, representatives from Italy, Denmark, and Portugal presented experiences in the adoption and use of AI in the public sector. Afterwards, a summary was presented of the feedback collected from the different breakout sessions around the recommendations.
The summary of the feedback was structured along with the four Intervention Areas in which recommendations are grouped. Five overall principles emerged in discussions across the different Intervention Areas and across the discussion groups:
- Avoid duplications where there is room for reuse of existing resources;
- The importance of training a wide array of stakeholders;
- Re-think procurement practices;
- Monitoring ex-ante and ex-post through observatories and experimenting via sandboxing;
- Refine the understanding of implications of co-creation.
The concluding remarks of the workshop highlighted the progress of AI Watch activities leading to the 4th Peer-Learning Workshop, and outlined directions for future initiatives at EU level concerning AI adoption and use in the Public Sector, and invited stakeholders to keep engagement high in order to progress on a mutually beneficial and complementary fashion.
Download the report here
Download the presentations here.
How to cite this report: Manzoni, M., Medaglia, R. and Tangi, L., AI Watch. Artificial Intelligence for the public sector. Report of the “4th Peer Learning Workshop on the use and impact of AI in public services”, 28 October 2021, Publications Office of the European Union, Luxembourg, 2021, ISBN 978-92-76-46347-4, doi:10.2760/142724, JRC127944.