Apply AI: GenAI for the public administrations -

Expected Outcome and Deliverables

• Pilots of European GenAI solutions in public administrations.
• Replication of piloted GenAI solutions across public administrations and Member States.
• GenAI4EU community of public administrations in Europe.

Objective

Generative Artificial Intelligence (GenAI) can play a transformative role in public administrations by automating routine tasks, enhancing personalized citizen-centric service delivery, and improving communication with citizens. It supports decision-making by analysing large datasets and simulating policy outcomes, while also helping in the drafting of legal texts. GenAI can also play a role in mitigating staff shortages and language barriers to access public services. Its implementation must take into account challenges related to privacy, robustness, sustainability, explainability, bias and transparency, while mitigating cybersecurity risks and ensuring human oversight in critical decisions complying with the AI Act

The objective of the call is to accelerate the adoption of GenAI in public administrations by supporting three to four pilot projects. Each pilot project will comprise one or more use cases where European GenAI solutions are developed and applied in the public administrations of the involved countries.

The pilot projects will focus on using European GenAI solutions to drive innovation, improve public services, and enhance citizen experiences, including through:

• supporting decision making in areas such as sustainable urban planning, infrastructure development, and transportation-systems design;
• optimising public administrations’ internal processes and operations, and promoting smarter budget planning and human-resource allocation;
• tailoring interactions with citizens via advanced platforms, such as chatbots and agents, to deliver personalized assistance, improve the accessibility of public services, and offer integrated support across multiple domains, including combining social protection, healthcare, social services, public employment, migration management, security, and other services into a seamless, one-stop experience; and/or
• making legislation more machine-readable (‘law as code’) and complicated procedures and text more understandable to citizens and businesses, for instance regarding environmental authorisations, procedures for starting a new business, or funding opportunities.

A key element of the piloted GenAI solutions will be their replicability across various EU public administrations, enabling consistent digital service delivery across Member States.

​​​​​​Scope

A call will be launched to select consortia of public administrations at national, regional, or local level that wish to participate in the pilot projects. The consortia may also include other entities such as higher education institutions, research and technology organisations, and civil-society and non-governmental organisations, which will support the public administrations in the implementation of the pilot projects.

To apply, consortia must submit a project proposal that includes:

• An overview of the planned GenAI models and solutions and the specific use cases where these solutions will be deployed across participating public administrations;
• An overview of the expected outcomes and benefits of the proposed GenAI models and solutions;
• An outline of the consortium's capacity to successfully integrate and scale up the proposed GenAI models and solutions; and
• A letter of commitment from the Member States or local or regional authorities of the participating public administrations, undertaking to provide 50% co-funding of the respective project costs, should the project be selected for funding.

Successful consortia will be awarded a grant to implement the proposed pilot projects. Specifically, public administrations within these consortia will be responsible for procuring the fine-tuning of proposed foundation models, developing tailored solutions based on these models, and integrating them into their existing platforms, systems, and operational workflows. The procurement could moreover cover supporting infrastructure and implementation activities such as:

• The technical infrastructure required to deploy and run GenAI solutions at scale and across single or multiple public administrations;
• The support needed by public administrations to ensure they have the necessary skills and knowledge to effectively implement and manage the deployed GenAI solutions; and
• The maintenance and optimisation costs for the deployed GenAI solutions.

Procured GenAI solutions should be designed to be compatible with various systems and platforms used by different public administrations to allow for easier replication across different contexts.

In addition, other consortium entities may provide support to the public administrations across a range of areas, including:

• Developing procurement specifications;
• Sharing knowledge and building capacities;
• Organising change-management and adoption-support actions;
• Documenting best practices; and
• Engaging and communicating with citizen and other stakeholders.

Entities that are part of a successful consortium will be ineligible to participate as suppliers in the procurement processes initiated by the public administrations within that consortium.

To ensure public trust, the procurement process will require the use of foundation models and systems which aligns with EU values and rules, particularly in terms of data protection, explainability, sustainability, and compliance with the EU AI Act. Projects that demonstrate EU-added value will be prioritized. This approach supports the EU's objective of fostering innovation while safeguarding fundamental rights and ensuring public trust in AI systems. As stated in the regulation establishing the Digital Europe Programme, the financial contribution from the Union should pursue as one operational objective under Specific Objective 2 – Artificial Intelligence: “build up and strengthen core AI capacities and knowledge in the Union”; therefore, solutions based on European models (i.e., models developed by European AI companies/laboratories) will be required. Overall, this action will support three to four pilot projects of the order of EUR 5-7 million of EU funding, which will be matched by an equal amount of funding by the project beneficiaries and or other public funding (e.g., from Member States’ national, regional authorities, etc.).

A separate call for proposals will be issued for a single Coordination and Support Action (CSA), with a budget of up to EUR 2 million. This call will be open to consortia comprising higher education institutions, research and technology organisations, civil-society organisations, non-governmental organisations, and other interested stakeholders. The successful consortium will be responsible for enhancing the scalability and replication of successful European GenAI pilot solutions, through activities that foster knowledge sharing, community building, and capacity development. Such activities could consist, for example, in implementing software documentation best practices, facilitating peer-to-peer knowledge sharing and experience exchange, deploying targeted training and support programs, and establishing a community of practice. These activities will ensure that the implementation strategies of successful pilot projects can be readily replicated, enabling seamless adoption across different public administrations and Member States, EEA-EFTA countries and countries associated to the Digital Europe Programme. This CSA will moreover support the creation of a GenAI4EU community of public administrations in Europe and its integration into the large European AI ecosystem of excellence. Participants in this action should in particular, cooperate closely with the European Digital Innovation Hubs and the AI-on-Demand-Platform, leveraging their expertise and building on their efforts targeted to the public administrations. Entities that are part of the successful CSA consortium will be excluded from participating as suppliers in the procurement of the public administrations involved in the pilot projects.

The action should establish links and build synergies with related initiatives, such as the Alliance for Language Technologies, the action on open-source European foundational model fine-tuning, the sectoral AI & Robotics Testing and Experimentation Facilities, data spaces and relevant EuroHPC initiatives. Furthermore, it should work with actions implementing the AI Act, such as the EU AI Innovation Accelerator and regulatory sandboxes, as well as with the AI Factories.

Strong links should also be built with the future Multi-Country Project on Innovative and Connected Public Administrations.

News flashes