The public needs input, and consensus, on how they should shape AI in their lives. Belgium’s Citizen Panel on AI promises much

Prompt to Midjourney, “voting in a digital world”

As partial fans of citizens assemblies’, set in a wider field of democratic innovation, we are intrigued to see a Citizens Panel on Artificial Intelligence being initiated by the Belgian government, as an element of their turn taking the Presidency of the European Union.

The Democracy Technologies website reports it thus:

A few weeks before the European Parliament’s final debate on the AI Act, 60 randomly selected members of the Belgian public convened in Brussels for a discussion of their own. The aim was not to debate a particular piece of legislation, but to help shape a European vision on the future of AI, drawing on the views, concerns, and ideas of the public. 

They were taking part in a citizens’ assembly on AI, held as part of Belgium’s presidency of the European Council. When Belgium assumed the presidency for six months beginning in January 2024, they announced they would be placing “special focus” on citizens’ participation. The citizen panel on AI is the largest of the scheduled participation projects. Over a total of three weekends, participants are deliberating on a range of topics including the impact of AI on work, education, and democracy. 

The assembly comes at a point in time with rising calls for more public inputs on the topic of AI. Some big tech firms have begun to respond with participation projects of their own. But this is the first time an EU institution has launched a consultation on the topic. The organisers hope it will pave the way for more to come.

The EU’s first citizens’ assembly on AI

Over the last two decades, the EU has been shaping a distinctively European response to the regulation of digital technology. It has established an international reputation for striking a balance between strong commitment to individual privacy rights and corporate responsibility with a desire to foster innovation.

The AI Act was an important step towards applying this to AI. Yet as the AI rollout continues, it is clear that this alone will not be enough. The technology is already having a direct impact on the lives of millions of people across Europe, and the long-term impact is hard to predict. A citizens’ assembly is one way of ensuring that future legislation is anchored in the concerns and needs of the public. 

The assembly has already met twice, on the weekends of 24 & 25 February and 23 & 24 March. The organisers left the scope of the discussion open, so that the participants could raise the topics they deemed important. In the first meeting, they established 6 areas of focus concerning the impact of AI: Environment, Defence and Security, Health, the place of the EU in the World, Work and Education, Democracy and Media. 

High levels of interest

The organisers report an enthusiastic response among the participants. Already at the planning stage, there were signs of high levels of public interest. Of the 16,200 people initially invited, over 1,000 registered an interest – a  large proportion for a sortition process. A group of 60 participants was then selected using stratified sampling to ensure that the final group was representative of the population as a whole. 

Beginning with the launch of OpenAI’s ChatGPT in November 2022, there has been an explosion of public interest in the topic of generative AI. International polls suggest that alongside curiosity and enthusiasm for the topic, there is also growing concern about its potential negative impacts.

The complexity of the technology and the vast scope of its potential impact presents legislators with a major challenges. For the organisers of the panel, this is one of the reasons why involving the public in the decision-making process is crucial.

Inès da Câmara Santa Clara Gomes, Attaché to the Belgian EU Presidency 2024, is one of the organisers of the citizens’ panel. She believes it is crucially important to include the public in discussions of AI.

“AI is a complex topic, of course, but it is on a lot of people’s minds”, she told us. “It’s already everywhere, it really impacts our daily lives. We have seen that the topic raises a lot of emotions. A lot of people expressed their fears relating to AI. Listening to these fears is important, as it helps to understand people’s social concerns about AI.”

More here. There’s an interesting take on this by Nathan Gardels at Noema Magazine. Gardels notes that the digital era, where individuals communicate with other individuals or in small social bubbles, attacks the classic idea of the public sphere, where we achieve a common understanding about issues:

…Following the insights of the philosopher Byung-Chul Han, democracy has largely given way to “infocracy” as peer-to-peer connectivity “redirects the flows of communication. Information is spread without forming a public sphere. It is produced in private spaces and distributed to private spaces. The web does not create a public.”

The possibility of arriving at a governing consensus through negotiation and compromise is being shattered by a cacophony of niche propagandists egging on their own siloed tribe of the faithful to engage in an endless partisan battle.

In short, the digital media ecosystem is disempowering the public sphere. Just as republics have historically created institutional checks and balances when too much power is concentrated in one place, so too we need to foster checks and balances for an age when power is so distributed that common ground can scarcely be found…

The broader implication of the Belgian experiment is a recognition that there can be other modes of advice and consent in today’s troubled democracies beyond the passive citizenship of periodically voting in elections dominated by partisan strife.

Indeed, it is becoming clear that democracy will only survive the fragmenting force of distributed communication technologies if it can find ways to bring the broader civil society into governance through common platforms that enable and encourage the search for social consensus.

More from Noema here. The final round of the Citizens Panel on AI is on 20-21 April, and the final report will be delivered on 25th May 2024. Here’s a highlights video of the March panel meeting