We The People’s first national conversation: Freedom and Equality
Earlier this summer, we announced We The People, a first-of-its-kind initiative in partnership with the Napolitan Institute that brings America into a nuanced, nationwide conversation, culminating at America’s 250th anniversary next year. Jigsaw and the Napolitan Institute are now launching our first national conversation, which will explore freedom and equality — concepts that have long-defined and continue to shape the American identity. We have invited thousands of Americans from every corner of the country to share their beliefs, opinions, and experiences on these topics.
Our goal is to help create a living, breathing portrait of what it means to be American today. To fully capture this conversation, we will use the latest Gemini AI models to analyze, distill and present the many perspectives shared, fostering a deeper understanding of Americans’ opinions on these topics.
How the conversation will work
For participants, the conversation will occur in three rounds:
Round 1: Individual Expression
The first phase of the conversation is designed to encourage personal reflections from participants. This begins with open-ended questions, such as “What does freedom mean to you?” Participants are encouraged to respond in their own words, without constraints on length or detail. This allows for a rich and uninhibited expression of individual perspectives. Following each initial freeform response, the survey employs Gemini to generate a tailored follow-up question. The goal of the follow-up question is to help participants go a level deeper in a live, personalized dialogue.
Round 2: Collective Reflection.
Subsequently, participants have the opportunity to explore and weigh in on the responses of other participants. In Round 2, we use AI to create an interactive set of opinions summarizing the responses from Round 1. Participants are invited to explore the opinions, and asked to respond to selected quotes and rank opinions. This helps us understand what discussion points were most important to the participants, and helps participants to learn what other Americans think, enabling them to see themselves in the national opinion landscape and feel part of a larger conversation.
Round 3: Conversation Validation
Based on the input from Rounds 1 and 2, our AI will generate declarative statements that are most likely to represent as many participant perspectives as possible. In the final stage, participants rate the statements based on how well they represent them and their ideals. This enables us to generate a final report that stackranks Americans’ ideas on the subject and where they agree or disagree.
Following the conversation, a summary report of findings and participant comments will be created and released publicly on wethepeople-250.org, a new site created by Napolitan Institute to host updates. Additionally, all anonymized participant responses, AI-generated follow up questions and large-language model (LLM) prompts will be published.
To ensure geographic representation across the U.S., The Napolitan Institute contracted with a third-party recruitment firm, Rep Data, to recruit over 2,500 participants from all 435 U.S. congressional districts. At a national level, we are aiming for a broadly representative sample of American citizens according to the 2024 Census. The recruiting methodology ensures that we are able to capture a holistic snapshot of perspectives while also having a measure of geographic representation across all congressional districts.
Key learnings from our pilot
Before launching with thousands of participants, we first piloted this process with a smaller sample of 200 participants to test and improve the methodology. Jigsaw is committed to open and transparent innovation, so we’re highlighting what worked, what didn’t and what we tweaked for the larger-scale conversation:
Tuning our AI to strike the right tone:
Public opinion research does not typically encourage deep reflection from participants at scale. Questions are either yes/no, or even when they are open-ended, respondents tend to share a few quick words that first come to mind. Using AI to generate follow-up questions invites participants to share more nuanced perspectives in a way that feels conversational.
One of the biggest challenges that emerged from the pilot was how to get the AI’s tone right, which has been a struggle across the industry. In early iterations, we heard from some participants that it felt too “artificial” or verbose. From others, we heard it was too direct, aggressive, or combative. We realized the original AI persona made the conversation in Round 1 feel like it was with a researcher: formal and, for some, unnatural. Ultimately, we developed a tone that made people feel like they were conversing with a curious and approachable conversation partner (like a neighbor or friend), who might surface deeper insights.
Transparency and clear instructions are also key. In particular, transparently disclosing when participants were interacting with AI, and being upfront about the purpose of the AI in the conversation are critical to building trust and keeping participants engaged in good faith.
In the end, we found that 94% of the pilot participants found the AI-generated follow-up questions helpful in exploring their ideas further.
“Personally, I came in with the expectation that the AI would be off in understanding my point and then would be asking questions that were either redundant or purely off base. However, the AI questions accurately understood my position and asked clear and direct questions that allowed me to elaborate and differ.”— Pilot Participant
An opinion map that invites exploration:
In Round 2, participants can see the views of others for the first time — an important step in any conversation. During the pilot, we found that while participants valued seeing a wide range of quotes, too many became overwhelming. To strike a balance for the large-scale conversation, we’ve implemented an interactive model that allows participants to explore a subset of varied quotes initially, while also providing them the option to easily click and explore a broader range of responses. We tested various models, anticipating that we would need complex visualizations of opinions in multi-dimensional space, but we heard instead that participants valued the simplicity of a list.
These visualizations enable participants to hear nuanced perspectives directly from fellow Americans — challenging oversimplified narratives and misperceptions. Three in four pilot participants gained a better understanding of different viewpoints, even if they disagreed with them. It’s an unprecedented opportunity to see where your own opinions fit into the nationwide landscape and to get a nuanced, interactive picture of that landscape.
Enabling the conversations of tomorrow
We The People is made possible by the next iteration of Jigsaw’s sensemaking technology — first used in Bowling Green, Kentucky in one of America’s largest-ever town halls. The technology uses Gemini to understand real-time conversation, capture nuance at scale and share meaningful insights in people’s own words.
At Jigsaw, we believe it’s more important than ever to build tools for people to join and stay in conversation. We are committed to building technology to give people voice and choice and believe this effort represents an important step in that direction. We invite you to embark on this journey with us. For the latest updates and insights from our inaugural discussion, please visit wethepeople-250.org.
