Center for Advanced Internet Studies (CAIS) gGmbH

A successful first Global Dialogue on AI Governance should lay the foundation for inclusive, practical, and forward-looking cooperation on artificial intelligence. At its core, success would mean a shared commitment to ensuring that everyone—not just a privileged few—can access, understand, and benefit from AI technologies. The growing gap between those who can effectively use AI and those who cannot risks deepening existing inequalities. Addressing this must be a central outcome. To achieve this, the Dialogue should prioritize three key pillars: infrastructure, competencies, and processes. First, robust and accessible digital infrastructure is essential to enable widespread and equitable use of AI systems. Second, building competencies at all levels—from governments and public administrations to local communities, and across all age groups—is critical. AI literacy should not be limited to experts; it must become a universal skill. Third, clear and well-designed processes are needed to integrate AI effectively into public services and governance. Another important outcome would be a shared understanding of how AI can enhance productivity and improve public services, while also identifying areas where automation of well-defined processes can create efficiencies. At the same time, the Dialogue should emphasize the importance of human oversight and accountability. Furthermore, success would include concrete steps toward participatory governance: better understanding AI users, enabling them to shape AI systems, and strengthening mechanisms for control and transparency. This should be supported by the development of ethical guidelines that foster trust and public acceptance. Finally, the Dialogue should result in commitments to invest in education, research—especially open-source initiatives—and pilot projects, alongside adaptive regulatory frameworks. Together, these outcomes would ensure that AI governance is not only effective, but also inclusive, democratic, and sustainable.

• AI capacity-building
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Open-source software, open data and open AI models

First, AI capacity-building is essential to ensure that individuals, institutions, and governments can effectively understand, use, and shape AI systems. Without broad-based competencies, the benefits of AI will remain unevenly distributed, reinforcing existing inequalities. Capacity-building must therefore take place at all levels-across public administration, political institutions, local communities, and among citizens of all ages. Second, we emphasize the importance of addressing the social, economic, ethical, cultural, linguistic, and technical implications of AI. AI systems are not neutral; they shape and are shaped by societal contexts. It is therefore critical to develop inclusive governance approaches that reflect diverse perspectives, safeguard fundamental rights, and promote transparency and accountability. Particular attention should be given to cultural and linguistic diversity to ensure that AI systems serve global populations fairly. Third, open-source software, open data, and open AI models are a key priority for fostering innovation, transparency, and accessibility. Open approaches can lower barriers to entry, enable public scrutiny, and support collaborative development across sectors and regions. They are especially important for public institutions and smaller actors who may otherwise lack the resources to engage with advanced AI technologies. Across these priorities, a unifying objective is to ensure that AI can be accessed, understood, and shaped by all. This requires not only technical solutions, but also investments in education, infrastructure, and participatory processes that empower users and strengthen democratic oversight.

First, the gap between AI users and non-users is an increasingly critical societal divide. While much of the current discourse focuses on access to AI technologies, less attention is paid to the ability to meaningfully use them. Differences in skills, confidence, and understanding create uneven benefits and risks. Addressing this "AI usage divide" requires not only infrastructure, but also user-centered design, education, and continuous support tailored to diverse populations. Second, human-AI interaction and usability should be treated as a core governance issue. Many challenges around transparency, accountability, and trust ultimately manifest at the interface between users and AI systems. My research shows that users often struggle to understand system behavior, limitations, and risks. Governance efforts should therefore include standards and best practices for explainability, interaction design, and feedback mechanisms that enable users to actively shape AI systems. We need to establish human-AI interaction as a distinct and visible field that systematically studies how people can meaningfully benefit from AI in real-world contexts. This field should examine not only usability and explainability, but also how AI can support human agency, learning, participation, and decision-making. It should further inform the development of teaching, training, and professional education so that people across sectors can learn how to work with AI critically, confidently, and productively. Third, participatory and inclusive AI governance remains underdeveloped. Beyond consultation, there is a need for mechanisms that allow users and affected communities to co-design, audit, and influence AI systems in meaningful ways. This includes integrating participatory approaches into public sector AI deployment and regulatory processes.

A central challenge is the growing divide between those who can effectively use AI and those who cannot. While AI tools are becoming widely available, meaningful use still requires skills, confidence, and institutional support. In public administration, for example, this leads to highly uneven adoption: some units experiment productively with AI (e.g., for text work, planning, or coding), while others lack the competencies or guidance to engage at all. This creates inefficiencies, missed opportunities, and risks of fragmented or uncoordinated deployment. Another key challenge lies in the lack of clear processes and standards for integrating AI into everyday workflows. Without practical guidance, many organizations struggle to move from experimentation to reliable, accountable use. Questions around transparency, oversight, and risk management often remain abstract and are not sufficiently translated into actionable practices at the user level. At the same time, there are significant opportunities. AI offers substantial productivity gains, particularly in knowledge work and administrative processes. It can support more responsive public services, automate well-defined tasks, and enable better decision-making if used appropriately. However, realizing these benefits depends on investing in capacity-building across all levels—from national institutions to local administrations, and across all age groups. Overall, the key governance gap is not only technological, but human-centered: we lack sufficient focus on how people actually use AI in practice. Addressing this gap—through education, infrastructure, and user-centered governance—will be critical to ensuring that AI benefits are broadly shared and effectively realized.

One important example is the Commission of Inquiry IV, "Artificial Intelligence – Toward a Smart State in a Digital Society," established by the State Parliament of North Rhine-Westphalia in Germany. This initiative brings together policymakers, researchers, and practitioners to explore how AI can be effectively and responsibly integrated into public administration and democratic processes. Its work highlights the importance of combining technological development with questions of infrastructure, competencies, and governance processes—an approach that could be scaled and shared internationally. In addition, the Dialogue should engage closely with established research communities such as ACM SIGCHI (Special Interest Group on Computer-Human Interaction). This global community has long-standing expertise in human-centered design, human-AI interaction, usability, and participatory approaches. It directly addresses many of the practical challenges around transparency, trust, and effective use of AI systems in real-world contexts.

Governments, international organizations, academia, industry, and civil society should be enabled to contribute written inputs, case studies, and policy proposals in a structured and comparable format. In particular, researchers and practitioners can provide evidence-based insights on real-world AI use, while public institutions can share experiences from implementation. A key recommendation for the format is to establish open online forums or consultation platforms, where experts and stakeholders can respond to guiding questions in a transparent and publicly visible way.

A key group that remains significantly underrepresented in global discussions on AI governance is everyday users of AI systems—that is, people who interact with AI in their daily work and lives, but are not part of technical, policy, or corporate expert communities. While current debates are often dominated by governments, industry leaders, and technical experts, the perspectives of users are critical because many governance challenges only become visible in real-world use. Issues such as misunderstanding system outputs, overreliance, lack of trust, or difficulties in integrating AI into workflows emerge at the point of interaction. Without systematically including users, governance risks being disconnected from actual practice.

Open online forums or consultation platforms, where experts and stakeholders can respond to guiding questions in a transparent and publicly visible way.