AI and the Commons: Participation in the AI Governance

During the last AI and the Commons call, we spoke to Tim Davies about including the public in AI governance. In his presentation, Tim talked about Connected by Data’s People’s Panel on AI, which took place during AI Fringe – a series of events organized to complement the UK Government’s AI Safety Summit.

AI Fringe was an effort to make the closed-doors summit a more inclusive event but ended up inviting mostly businesses and nonprofits, leaving the broader public “outside.” The People’s Panel, a deliberative review of both the UK AI Safety Summit and AI Fringe, was meant to broaden the participatory scope of these events. This made the learnings from the panel particularly interesting for us – Open Future has also recently launched its own deliberative participation process, the Alignment Assembly on AI and the Commons.

Deliberative participation: what does it take?

People’s Panel on AI brought together a representative mini-public of people from across England to spend four days learning about AI, engaging with AI experts, and discussing their thoughts. Deliberative participation focuses on discussion and debate with an aim to reach a consensus. While thinner participation techniques such as polls and surveys can help reach a bigger group of people, deliberative participation is a thicker, more intense group process. Through a facilitated deliberation process, the Panel ended up putting forward recommendations for government, industry, academia, and civil society.

To make sure the selected group is representative of the UK society (with a higher target for ethnic minority representation and, as it happens, a higher target for retirees as they were the ones available to join offline events at short notice), Connected by Data relied on the help of the Sortition Foundation to come up with a sample participants group. After the selection process, the People’s Panel on AI ended with a final group of 11 participants. They were first contacted by phone and email (depending on the level of digital literacy) and then onboarded via an online webinar, which helped set the scene for the physical in-person sessions. This introductory process clarified the next steps and created a shared understanding – a common ground needed for a free exchange of opinions.

While designing and executing this project seems (and definitely was) a lot of work, it actually only took six weeks from design to delivery, making it a very lean and potentially reusable strategy. However, the success of the model depended on some critical infrastructure made available both by the Sortition Foundation’s methodology and tools, and – very importantly – AI Fringe budget and set-up. You can read more about this case study in a report here.

Why participation matters: how to set up expectations?

The recommendations of the People’s Panel on AI were presented at AI Fringe and captured in the news around the event. The participants of the panel subsequently got invited to join the TechUK Digital Ethics Summit as well as other local events and conferences on the topic. Hopefully, they can continue being seen as a valuable asset and kept in the loop, even though that usually calls for ongoing coordination and funding.

It is hard to imagine a direct, political impact of all the recommendations and principles defined over the course of citizen assemblies and virtual deliberative processes (such as Collective Intelligence Project’s Alignment Assemblies on AI and our own) organized around the world.

However, this rise in participation on AI issues is a direct response to the growing power of big tech and the criti-hype surrounding AI. It marks how people and communities are trying to regain influence.

There is value in creating (and holding) a space for connection and overlap between powerholders, the public, and experts. During the talk, Tim used an example of how a debate between policymakers or experts can be affected simply by allowing a different audience in the room: it changes the nature of the arguments used.

Creating opportunities for including the public in these processes can resurface a different narrative by reminding everyone about the public interest and to whom they (and technology) should be ultimately accountable. The insights generated during participation processes can also get picked up by the media and help shift the current debate about AI, demonstrating that it is not just the usual suspects, the establishment, that have the voice.

As testimonies of the People’s Panel participants remind us, well-designed participation processes empower “the public”— people who usually aren’t a part of these conversations – and encourage them to take part in similar conversations, and the democratic process, in general. Participative deliberation projects can help set standards for regulating and governing technologies as the Commons: in the public interest and with people’s engagement and oversight.

AI and the Commons community calls are invite-only conversations. If you’d like to join them, email alicja@openfuture.eu.