Incite’s Trust Collaboratory Hosts The First Community Conversation for AI Research Project - Incite at Columbia University

The specifics of the development of AI technologies is top-down and developed without public input, bypassing the needs for democratic solutions to the public’s needs and priorities. But what could an AI look like that people actually wanted to use, and what processes for intellectual engagement, co-design, and development would we need to build a more democratic AI?

The first gathering of an AI community conversation for the Trust Collaboratory’s cooperative research project, TeLLLMe, convened on Monday, June 29 to create a dialogue between community members and researchers to tackle these questions. 

The Trust Collaboratory’s Director Gil Eyal and Associate Director Cristian Capotescu are social scientists researching trust in AI to aid the creation of a publically useful AI platform built on accurate public information. Together with engineers and data scientists from Columbia University and Rutgers University, the TeLLLMe team is currently working to create a prototype that serves the common good. 

“Our challenge as a project is to build a community of practice around this project and to work with community partners,” said Capotesu. “We want to open the door to non-users of AI, and listen carefully to people that are very skeptical of AI. And hear, what does skepticism and mistrust mean?”

In addition to non-users, the TeLLLMe team is also opening their doors to tech enthusiasts and everyone who does not fit neatly into a positive or negative view on AI development. For the researchers, a diverse group of participants creates a variety of different positions on their topic. 

The project is funded partially by the National Science Foundation and over the project’s timeline from 2025 to 2027 is invested in defining what makes AI accountable, fair, safe, transparent and trusted. 

The community conversation included researchers, teachers, students, and entrepreneurs. The group sat around a table, engaging respectfully as the group voiced their varied opinions. 

The call for participants began in Harlem, but welcomed people from across New York City to participate. The resulting conversations created an open dialogue that flowed easily between questions and answers from both researchers and community members. 

“It's not good for trust in technology if you deploy a product that is full of errors and offers misleading information,” said Capotescu, emphasizing that when corporations or AI developers iterate on technology in public without taking into account users' needs and priorities before they release it, this can be highly corrosive to trust. “In order to prevent this from happening, you have to start with listening to users, listening to what their needs are, providing guard rails, having a deliberate process, going slow, continuously listening to public opinion, and integrating public stakeholders into the development of tools that are placed in consequential roles in society.”

The gathered group of social scientists, researchers, and artists described this initial meeting as “inaugural and lowkey,” but the response from the gathered group was enthusiastic about the idea of continuing to meet throughout the summer and for the foreseeable future. One participant emphasized that she had observed that the people across New York City increasingly invested in meeting to discuss the technology that has been seeping into their everyday lives.

“So many different communities are gathering to discuss AI and it's all in person,” she said.

Alongside the promise of future listening sessions, the conversation closed with a participant’s simple statement: “Trust is necessary for me.” Her words align with the Trust Collaboratory’s work to understand how trust and democracy can be repaired, can intersect, and can build off of each other; and how these ideas can be explored through listening and dialogic formats. 

The Citizen AI Project’s future events will be shared on Luma, or can be explored on the Trust Collaboratory website.