AI is testing democracy’s capacity to govern

Artificial intelligence is becoming a test of political institutions’ ability to act. As its uses expand, governments face decisions whose consequences extend well beyond the technology sector. How much authority should be delegated to automated systems? Who should be accountable when they cause harm? How should the benefits of AI be shared?

At the Athens Democracy Forum, held from 29 September to 1 October(1), these concerns featured in a conversation between Kyriakos Pierrakakis, Greece’s Minister of National Economy and Finance, and Thomas L. Friedman, a columnist at The New York Times. Moderated by journalist Patrick Healy, their discussion focused on the gap between the pace of technological change and the capacity of political institutions to respond.

“Institutions are not fit for purpose for the AI period. Our institutions were built for the previous era,” Pierrakakis argued. His diagnosis was that technology is changing exponentially, while our perception of change remains linear and political processes move more slowly still.

Friedman approached the same problem through a metaphor: technological breakthroughs can change society’s state, much as heat turns ice into water and water into vapour. In his account, AI marks a transition whose scale and consequences we have yet to grasp.

The political difficulty is clear. Governments must make decisions while the technology, its uses and our understanding of its effects are still evolving. Waiting for certainty has consequences; acting quickly also means making choices with incomplete knowledge.

For democracies, speed cannot be the only measure of a good response. Decisions must remain open to scrutiny and challenge. People need ways to influence them, and institutions need the capacity to revise them as evidence and experience accumulate. The question is how to respond to rapid change while maintaining those democratic requirements.

The discussion in Athens matters because it brings this problem into political view. A serving minister was asking whether the institutions through which we govern are equipped for the changes AI brings. That question concerns the functioning of democracy itself.

Global rules, and a public role in shaping them

Pierrakakis and Friedman agreed that no country can govern AI alone. The minister compared its development to both the race to the Moon and the nuclear arms race: a source of opportunity, but also of competition and risk. Friedman called for the United States and China to work towards common rules.

International cooperation is essential to that response. It also leaves a democratic question unresolved: how can people influence the choices being negotiated on their behalf?

Governments and technology companies have resources and authority to shape AI’s development. Citizens experience its effects in their workplaces, in education and in their dealings with public services. Their priorities may differ according to where they live, the opportunities available to them and the risks they face.

For Missions Publiques, these experiences should help inform decisions about AI. Asking people whether they support or fear the technology tells us relatively little about the choices they would make after examining its possible uses and consequences.

Citizen deliberation allows a more substantial discussion. Participants have time to learn, question experts and consider competing arguments. They can examine specific trade-offs: under what conditions should a public authority use AI to make or support decisions? What opportunities should people have to contest those decisions? What protections would they expect before accepting a particular use?

Such discussions can reveal areas of agreement, but also disagreements that decision-makers need to understand. They can show which safeguards people consider necessary and why a proposal that appears effective to its designers may be unacceptable to those affected. To contribute to governance, however, this work needs a route into decision-making. Participants must know who will receive their recommendations and how those institutions will respond.

This is the ambition behind the Global Coalition for Inclusive AI Governance, launched in February 2025 by Missions Publiques and the Stanford Deliberative Democracy Lab at the AI Action Summit in Paris, with support from France, Switzerland and the United Kingdom.

The Coalition aims to involve more than 10,000 citizens in over 100 countries through decentralised citizens’ assemblies and deliberative polls. Their recommendations would be brought to decision-makers working on AI governance.

The international scope matters. AI is being developed and deployed across borders, while opportunities to influence its direction remain uneven. Bringing deliberation to different countries would allow people with different experiences to examine common questions. Comparing their conclusions could help identify shared expectations and differences that global rules need to take into account.

Missions Publiques brings experience from earlier global deliberations, including World Wide Views on Climate and Energy ahead of COP21 in 2015, and We, the Internet in 2020. These processes showed that people from varied backgrounds could engage with complex international issues when given time, accessible information and support to deliberate.

The Coalition now brings together more than 80 civil society organisations and private sector actors, alongside academic partners. An initial pilot took place during the February 2025 summit, with an assembly of 70 randomly selected students, coordinated with the École normale supérieure (ENS-PSL) and Yale University. The next objective is to launch a first large-scale cycle of deliberation and establish links with decisions at national, regional and international levels.

The conversation in Athens raised the question of whether democratic institutions can keep pace with AI. Part of the answer will depend on their ability to draw on the judgement of the people affected and to explain how that judgement informs their decisions.

To support the initiative or join the Coalition, visit global-ai-dialogue.org.

(1) Missions Publiques at the Athens Democracy Forum

The Athens Democracy Forum is organised by the Democracy & Culture Foundation in partnership with The New York Times. Yves Mathieu, co-founder of Missions Publiques and a member of the Foundation’s Board of Governors, attended the 2026 edition. Further reading / Athens Democracy Forum / Watch “A.I. vs. the World”, with Thomas L. Friedman and Kyriakos Pierrakakis