WIREDGOV

It’s the million euro question: can AI-powered tech help scale public deliberation so that many more citizens can play a part in shaping the decisions that affect their lives?

We hope to answer that question with Waves, a €1 million digital democracy trial delivered by Demos, alongside a coalition of partners including New Local, CASM Technology, digital democracy tools, and Camden and South Staffordshire Councils.

Our first trial is live in Camden. We had over 1,200 Camden residents take part in phase 1, which was all about getting as many people as possible across the borough to answer a simple question on adult social care: “If you or a loved one needed care and support in the future, what would matter most?” This gave us resident priorities to take into phase 2, where 41 of those residents – randomly selected to be reflective of Camden’s population – took part in a digital deliberative process. Participants were provided with the time, space, and information to dig deeper into the topic of adult social care over the course of three online sessions.

Phase 3 is now live, where we’re going back out to wider Camden residents to see what they think of the ideas that have come out of the conversation so far – if you’re a Camden resident, you can take part here. In phase 4, we’ll then go back into deep deliberation with the smaller group of residents to make sense of what comes out of phase 3, and come to some final conclusions.

I’m going to share my own reflections on phase 2 on how you avoid losing the human as you scale deep deliberation through tech, looking at the impact on two key human aspects of deliberation: building relationships between participants, and facilitating good deliberation.

Building relationships between participants

People talk about the magic of deliberation. It’s what happens when you put ordinary people together in a room, people who wouldn’t otherwise have spoken to each other because of their differing life experiences or views, and get them to work towards a shared goal of reaching consensus. Participants make genuine friends, change their minds on things they never thought they would, and grasp complex ideas and trade offs on an issue they previously knew nothing about.

But it’s not magical really, it’s human, and relationships are at the heart of it. We only think it’s magical because we’ve lost the belief that it’s possible. And if it’s human, do we lose this when we bring tech into the picture?

In phase 2, we used a combination of the tech platforms PSi and Zoom to hold our deliberative sessions. Feedback was overwhelmingly positive from participants about the sessions, and we could see them bonding with each other. One aspect that was new was holding part of the conversation as audio only, so participants couldn’t see each other. Would this help participants feel comfortable speaking up, or hinder relationship building?

Starting the conversations with videos on was crucial for relationship building, as participants could then remember and relate to each other once they were in the audio only rooms. In fact, several participants mentioned themselves that they would’ve preferred to be able to see each other in those discussions. And as I’ll touch on below, audio only didn’t seem to be the key ingredient to encouraging participants to speak up.

Learning: Encouraging participants to have their videos on is key for relationship building to create the ‘magic of deliberation’.

Facilitating good deliberationOne of the other key human elements of traditional deliberation are facilitators, there to structure the conversation and keep time, ensuring everyone gets a chance to speak and the discussion moves forward constructively. We’ve all been in conversations where we patiently (or impatiently) listen as someone monologues to the group. We can’t harness collective intelligence if we’re only hearing from the loudest voices.

That’s why I was sceptical about not having a human facilitator in each small group discussion, which is one of the ways that tech could help scale deliberation. Tech enables essentially limitless numbers of participants to be in small group discussions at the same time, but you’re far more limited by capacity and budgets if you need a facilitator in each small group discussion.

We tested both ways – part of the conversation was facilitated by a trained facilitator, part of it was largely self-facilitated by participants, who were virtually nudged if they hadn’t spoken. We found that while initially participants struggled to self-facilitate and some were even sitting in silence when we listened in, they increasingly got the hang of it over time. The virtual tables were frequently mixed up, which brought new energy into the room, and we heard participants bringing in ideas from their previous tables without our prompting. In the end, the vast majority of participants felt they had opportunities to share their opinions and ideas during the sessions where there was self-facilitation, with 83-93% agreeing with this statement across the sessions. 

While I was impressed by some of the self-facilitated conversations I listened into (who knew participants could possibly have a conversation without the key ingredient: ME), there were still plenty of silences, particularly from quieter participants. For example, one participant said: “Two participants hardly said a word, not because they didn’t have anything to say but they seemed more than a bit nervous”. Also, several participants told us they felt the self-facilitated discussions were too long, even though they only lasted 10-15 minutes – it’s very rare for a conversation to dry up in that space of time with a trained facilitator there to prompt and guide the conversation, and ensure that quieter voices have their say. 

Learning: The jury’s still out on going without human facilitators in small group discussions, but it’s clear that there’s more we can do to prepare participants if we’re asking them to self-facilitate, and more that the tech can do to play the facilitator role by prompting participants in clever and constructive ways.

Answering the million euro question

To come back to the million euro question: can AI-powered tech help scale public deliberation? Yes it can, but there’s a lot we (and tech developers) need to think carefully about to not lose the magic of deep deliberation along the way. We’ll keep you updated on the magic as we test more tech tools and kick off our new trial in South Staffordshire.