Loading...
About
Talk.
Listen.
Be heard.
realtalk@Boston is a collaboration between the MIT Center for Constructive Communication, Cortico, and a growing network of local community organizations in Boston.
Together, we launched a new civic initiative designed to amplify the voices of Boston community members who are often underheard by current civic processes.
11
community organizations
50
hours of conversation
253
local participants
47
conversation themes
About the MIT Center for Constructive Communication (CCC)
Our mission at CCC is to design tools, methods and systems that connect rather than divide us — helping build a healthier, more empathetic society. Based at the MIT Media Lab and working closely with the non-profit Cortico, CCC brings together researchers, designers, technologists, and community partners across disciplines to explore how technology can strengthen human connection.
realtalk@Boston invites Bostonians to slow down and share real stories about life in our city — what feels hard, what feels hopeful, and what we want to build together as we imagine the future of Boston. Led by the translational research team at CCC and a cohort of local community partners, this citywide listening experiment uses new tools to connect the stories of Boston’s people into a collective portrait of belonging and possibility.
Process
Our approach begins by inviting small groups of people into conversations facilitated by local community organizers. Each conversation is recorded, and focused around the same key questions:
- What brought you to Boston?
- What are some supports you have found here? What do you worry about on a day-to-day basis?
- What do you hope for the future of Boston 10 years from now?
These recordings are the source material for sensemaking: a process to make meaning from collective experiences.
To understand these conversations as a collection, we put human listening first, then used AI and natural language processing (NLP) tools to support synthesis.
Methods and Tools
Working closely with the non-profit, Cortico, we use their platform to review conversation transcripts and, as a team, highlight the moments that resonate most deeply. These often include personal stories, ideas, and insights that reveal something meaningful about community life.
Once those moments are gathered, AI tools help us look across many conversations, tracing patterns and connections that begin to tell a larger story about what matters to the communities we are working with
To find patterns across communities, we collaborate with an AI model to “tag” highlights with themes. During this process, conversation highlights are input into Large Language Models (LLMs) which suggest recurring themes or “tags”. Next, human reviewers refine these suggestions to create a thoughtful, organized database of themes and highlights.
Paper: Kabbara et al. (2025)
We use Large Language Models (LLMs) to summarize content within and across conversations, across themes, and organizations. These summaries are used throughout the sensemaking process as a way to bring in more context to highlights.
Paper: Mohanty et al. (2024)
We ask more complex questions about our data using Retrieval Augmented Generation (RAG) search. This is a smart search engine paired with a Large-Language Model (LLM) that finds, cites, and analyzes relevant quotes from transcripts in response to user queries.
To define narratives that combine themes into richer stories about life in Boston, we enter multiple themes into the RAG-search engine, seeking for both positive and negative instances in conversations.
Paper: Schroeder et al. (2025)
Contact Us
realtalk is an ongoing program seeking partners and collaborators. If you are interested in the work, please reach out to us at realtalk@mit.edu, or sign up for the MIT CCC newsletter below to keep up to date on the latest realtalk work.
Credits
Marina Rakhilin
realtalk@Boston Program Manager
Cassie Lee
Design and Creative Direction
Maya Detwiller
Prototype Manager
Lucas Drummond
Design and Software Development
Artemisia Luk
Photography and Social Media
Dimitra Dimitrakopoulou
Head of Translational Research
Funding
Support for this program was provided by the Robert Wood Johnson Foundation. The views expressed here do not necessarily reflect the views of the Robert Wood Johnson Foundation.