Governance framework, policy roadmap and recommendations
General Overview & Rationale
The framework for democratic AI governance and the associated roadmap and recommendations (KER 2) seek to foster regulatory responses to ensure AI systems do not undermine but empower democratic participation. The centralization of power, opacity and the speed of change inherent to current AI advances all generate risks to democratic practices. The framework details the harms and benefits AI poses to different pillars of democracy (e.g., participation, knowledge, freedom, equality) as well as how different AI governance approaches map against these pillars.
The aims of this KER include:
- Create a better conceptual understanding of the risks and opportunities the rapid deployment of AI systems in society poses for democracy
- Outline a roadmap for more inclusive and democratic AI policies in the future by engaging European policymakers, researchers and CSOs
- Provide recommendations on democratic governance and regulation of AI systems to protect citizens’ fundamental rights
The research also examines the feasibility and gaps of the established legislation such as GDPR and the AI Act in ensuring democratic use and regulation of knowledge technologies. It will not only consider risks like algorithmic bias and disinformation, but also the positive potential of AI and big data to reinforce democratic practices. It seeks to provide easily applicable guidelines to assessing algorithmic risks to democracy in the context of European AI regulation. In contrast to existing trustworthy AI governance frameworks, KT4D specifically focuses on the democratic barriers and opportunities of AI systems.
Input from Use Cases
The governance framework, roadmap and recommendations for democratic use of knowledge technologies will be developed iteratively with Use Case 1. Three co-creation roundtables with policymakers working on AI governance are organised in Brussels. The first meeting, held in October 2023, focused on identifying the blind spots and gaps of current regulatory approach to AI in Europe, especially with respect to democratic values.
Moreover, the governance framework will be based on a policy Delphi study on the future of AI governance conducted with policymakers from April to June 2024. The two-round policy Delphi study brought together twenty experts in AI governance and policy. They assessed the likelihood and desirability of different forward-looking AI policy options to explore consensus and disagreement surrounding them.
Findings & Next Steps
The first draft of the governance framework will be finalised in December 2024. It will be complemented by a policy roadmap and recommendations at the end of the project in January 2026. In the meanwhile, the framework will be updated and validated through the use case 1 meetings in Brussels with policymakers.
Stakeholder Groups:
Members involved in the development process
Stakeholders & Testing
The framework for democratic AI governance is mainly directed towards policymakers and associated stakeholders, such as CSOs working on AI policy. The Use Case 1 in Brussels has engaged European policymakers from the European Commission and Parliament, researchers from universities, think tanks as well as NGO representatives. Moreover, the Delphi study allows for identifying mutual priorities and different consequences across the stakeholders, making it valuable for navigating emerging issues of AI governance. It will be developed further iteratively in Use Case 1 meetings. The KER 2 also seeks to enhance democratic participation and citizen input into the regulation of AI systems. As such, citizens are also an important stakeholder group whose fundamental rights are protected through the framework.
Useful Resources
Why is it important?
The governance framework, policy roadmap and recommendations will bring together the key findings of the KT4D project into practical governance and regulatory guidelines for policymakers audiences. It provides guidance on how the EU should regulate the disruptive nature of general-purpose AI systems in service of democratic values.
----------
“The rapid advances in AI systems and the centralization of power to few dominant companies require urgent policy responses. The governance framework developed in KT4D helps in conceptualising and enacting more democratic forms of AI governance in response.” - Atte Ojanen, Demos Helsinki