Technology is rapidly changing facilitation, opening new horizons of scale & speed while also creating new risks for trust & democratic legitimacy. Many practitioners are experimenting with AI tools in their work, while technologists & developers are building LLM-powered features intended to scale the effects of facilitation. Yet not all components of facilitation or discussion are well-suited to AI support. And introducing new technology into sensitive discussions, especially in communities affected by conflict, is unlikely to succeed without first addressing concerns about bias, errors, privacy, & trust.

This report evaluates AI’s capabilities alongside critical guardrails found in the lived experiences of facilitators. It aims to inspire new AI features that meaningfully help facilitators rather than hinder them, augmenting the work rather than replacing it. Amplifying the voices of expert facilitators from around the globe, it is meant to complement the work being done by other researchers, developers, practitioners, & funders to enhance collective dialogues in the age of AI.

Across contexts, we found that structured dialogue processes are categorized by 3 different primary objectives. These objectives are really ideal types, highlighting key distinguishing features that set one kind of conversation apart from another:

• Transformative. Focused on affecting participants as individuals or as a group, e.g. learning, growth, empathy or capacity-building.

• Generative. Focused on gathering or generating a wide range of ideas, &/or mapping the opinion landscape in higher nuance & fidelity.

• Deliberative. Focused on coming to agreement &/or decision together through active discussion & learning.

Each objective is met by combining different process elements (e.g. public input, in-person discussion, expert presentations, etc.). The same set of activities is therefore not required for every structured conversation. In fact, elements not aligned with the main objective can feel unnecessary or undermine group trust.

We identified 5 opportunity areas where technology can best support and enhance collective dialogues, grounded in the needs and advice of expert facilitators:

• Scaled Access, via multilingual communications & tailored outreach tools. AI to reduce language & access barriers to participation in dialogues.

• Dynamic Learning, via multimodal education & evidence features. AI to make learning fully customizable & more engaging.

• Live Sensemaking, via real time conversation analysis & visualization tools. AI to help facilitators understand conversation dynamics in real-time.

• Futurecasting, via scenario planning & tradeoff extrapolation tools. AI to help people grapple with the risks & benefits of the decisions they’re making.

• Sensemaking for the People, via interactive public analysis & reporting tools. AI to give participants more power to shape reporting & impact.

Built collaboratively by facilitators, technologists, researchers, & publics, these 5 AI technologies could enhance conversations in meaningful & inspiring ways.

• Ian Beacock (Jigsaw/Google)

• Emily Saltz (Independent)

• Wasim Almasri (ALLMEP)

• Mahmoud Bastati (Build Up)

• Alessandra Cardaci (Debating Europe / Friends of Europe)

• Albert Cevallos (CANVAS)

• Cui Jia Wei (vTaiwan)

• Caleb Gichuhi (Build Up)

• Andrea Gallagher (Staple & Spindle)

• Beth Goldberg (Jigsaw/Google)

• Michele Holt-Shannon (University of New Hampshire)

• Nicole Hunter (MosaicLab)

• Emily Jenke (DemocracyCo)

• Felix Kufus (CMI)

• Scott Lappan-Newton (Gauge)

• Thea Mann (Jigsaw/Google)

• Alice Siu (Stanford University)