AI regulation cannot be left solely to technicians, politicians or companies. Decisions about how AI is used in society concern everyone: the worker whose CV is filtered by an algorithm, the patient whose treatment is influenced by a predictive system, the citizen whose neighbourhood is monitored by cameras with facial recognition. Yet these people are almost always excluded from the decision-making process. Participatory governance proposes to change this dynamic.
Citizen assemblies are the most promising tool. A group of citizens, selected by lottery to be representative of the population by age, gender, education and origin, is convened to deliberate on a complex issue. Participants receive information from experts with diverse perspectives, discuss in facilitated small groups and formulate recommendations. It is neither direct democracy nor a poll: it is informed deliberation.
Ireland used citizen assemblies to address divisive issues like same-sex marriage and abortion — achieving results that parliament had failed to reach in decades. France used the Convention Citoyenne pour le Climat to develop 149 proposals on climate change. Taiwan regularly uses the Polis platform for digital deliberation on technological issues, allowing millions of citizens to express preferences and find points of consensus.
Applying these tools to AI is a still-young idea but with concrete examples. The Ada Lovelace Institute in the UK organised citizen assemblies on facial recognition, producing recommendations that influenced the parliamentary debate. The EU's ALGO:AWARE project experimented with public deliberation on algorithms. The city of Amsterdam created a public register of algorithms used by the municipal administration, an example of transparency that allows citizens to know — and therefore to challenge — how AI affects them.
But participation must not be merely consultative: it must have real power. A citizen assembly whose recommendations are ignored is worse than no assembly — it creates cynicism and distrust. This is why mechanisms are needed to ensure citizens' recommendations are taken seriously: obligation for public response, implementation commitments, periodic review.
Participatory AI governance is not utopia: it is a democratic necessity. Technologies that shape our society must be governed by society — not only by those who develop them or those who regulate them, but by all of us. Relatronica believes this is possible and works to make it real.