Using AI to Scale Deliberative Engagement

This is the second of three articles appearing here on how artificial intelligence can scale public deliberation processes to include large numbers of participants, and why this is important for future governance.

Our first article in this series identified a Catch-22 in deliberative mini-publics: they need government backing to gain public support but can’t win public support without government backing. This second article explains how Artificial Intelligence (AI) lets us scale deliberation to solve the Catch-22.

Scaling Deliberation

Traditional consultation processes focus on collecting people's views and often include thousands of people. By contrast, deliberative processes go deeper, asking participants to discuss and consolidate their views, which presents a technical challenge: As the number of participants increases, so does the volume of qualitative and quantitative data generated. Using conventional tools to collect, analyze, and manage this data quickly becomes too labor-intensive and costly to be effective. Deliberative processes therefore typically are limited to about 25 – 50 people.

AI is a game-changer. It introduces a new generation of tools for managing and analysing qualitative data efficiently and effectively, making deliberation possible with very large groups. However, to succeed, the right technology and the right process must be combined in the right way. Let’s examine how this works, starting with the technology.

Leveraging AI to Support Deliberation

To effectively manage and analyze the data from large-scale deliberative processes, AI relies on a few key technologies, including:

• Text Summarization: Creates concise summaries.
• Dialogue Analysis: Understands conversational context, sentiment, and intent, identifying key insights.
• Sentiment Analysis: Gauges community emotions, detects agreement and contention, enabling real-time adjustments.
• Topic Modeling: Identifies underlying themes and trends, categorizing feedback and prioritizing actions.

As we will see in the next section, with the right process, these tools can work together to summarize and consolidate insights from the dialogue, enabling it to scale effectively and inclusively. So, what is the “right process?

Getting the Process Right

The Public Dialogue

Imagine a public dialogue process divided into three stages: Defining the Issues, Finding Solutions, and Validation. Each stage includes multiple events like online surveys and forums, townhall meetings, and community gatherings, engaging thousands of people.

Throughout the process, participants follow basic rules of respectful listening and give-and-take to work toward shared solutions. Events can be professionally facilitated or self-organized using official toolkits. Interventions are recorded and submitted to process managers, who then use the AI tools (see above) to analyze and consolidate the participants’ views and exchanges. This technology can then distill and surface the issues, insights, and emerging trends, enabling administrators to rapidly compile and publish them in updates. At the end of each stage, the results are consolidated and published in a findings report (see next section).

AI thus takes a data management task that would be unmanageably complex for conventional tools and transforms it into a very manageable one that allows the dialogue to scale effectively and inclusively. The process ensures that participants receive timely feedback on what is being said at other events, allowing them to consider and react to developments in the dialogue as it unfolds. As a result, participants are drawn into an increasingly focused discussion, which evolves and aligns organically and at scale. Nevertheless, a measure of leadership is still needed to facilitate this convergence, as the next section shows.

The Deliberative Group (DG)

Alongside the Public Dialogue, let’s now introduce a mini-public of about 20 people, which we’ll call the Deliberative Group (DG). The DG reviews, assesses, and consolidates findings from each stage of the Public Dialogue and makes final recommendations to the government. Basically, it does the deeper deliberative dive needed to ensure that each stage achieves its goal. The DG thus provides a measure of leadership for the public dialogue.

The DG's process includes four phases and runs parallel to the public dialogues, starting slightly before and ending slightly after each stage. Initially, DG members are briefed on issues, options, and likely views. They observe and review findings, then deliver reports setting the stage for subsequent discussions. This process repeats for each stage, with the final report providing recommendations to the government.

Throughout this process, the DG’s deliberations support and reinforce the public dialogues by elaborating and consolidating the findings, which are released in a report at the end of each phase, then reviewed by dialogue participants in subsequent stages of the public dialogue.

Standing back, we can see that these two engagement streams—the Public Dialogue and the Deliberative Group—are interactive, iterative, and mutually supporting. Each stage of the Public Dialogue provides material for the DG’s deliberations, which in turn provide material for the next stage of the Public Dialogue. This back-and-forth refines and consolidates the findings, aligning the Public Dialogue and the DG, and resulting in a growing sense of mutual ownership among all participants.

Solving the Catch-22

Summing up, our proposed process produces three critical outcomes that overcome the Catch-22 in existing deliberative mini-publics:

1. Engaging large numbers of people builds broad-based ownership of the results. AI tools can process large volumes of data reliably and quickly, ensuring that voices from all parts of the process get heard and considered. The combined insights provide a comprehensive assessment for the Deliberative Group’s deliberations.
2. Ownership mobilizes participants behind the results. Rapid feedback allows participants to respond to the Deliberative Group’s work, helping to shift and align their viewpoints on the issues and emerging solutions. Summaries show how their views have evolved and contributed to the collective outcomes, enhancing their sense of ownership and commitment.
3. Mobilization attracts government interest and support. Governments will be interested in scaled deliberative engagement because it addresses a serious problem. If a minister wants to implement a policy but fears that the opposition can successfully mobilize opinion against it, they need to build broad public ownership of the policy to counter these forces and create momentum. Scaled deliberative engagement can achieve this. If the recommendations are acceptable, it’s a price the minister will willingly pay for success.

Producing these three outcomes through AI-enabled, scaled deliberative engagement solves the Catch-22 for mini-publics. AI lets us effectively engage large numbers of people on an issue, where previously the demands on time and resources made this impossible. This large-scale deliberation results in broad public ownership and support for the results, which in turn wins government backing. But can we be confident that government will find the recommendations from such a process acceptable? Yes, and we’ll explain why in our third and final article.

This article was written by Don Lenihan PhD, Public Engagement Expert, President and CEO of Middle Ground Policy Research, Damian Carmichael former public servant with the Australian Government and Engagement Lead at Converlens and Tom Workman Director Converlens