realtalk@MIT brings together human conversation with digital technologies to foster authentic and nuanced dialogue across the MIT campus. Developed in collaboration with CCC’s closely affiliated non-profit Cortico as an extension of a previous on-campus pilot program, realtalk enables the MIT community to come together in facilitated, small-group conversations to share stories about common experiences and interests. Participants consent to having their conversations recorded – and again to have their audio highlights shared – so that they may elevate one another’s stories for their peers and our community to hear. Human-steered AI tools will help trained “sensemakers” to surface patterns and themes across these stories, while audio medleys, data visualizations and other outputs will help us all listen and learn from this remarkable constellation of voices. 

Starting this fall, incoming undergraduate and graduate students will talk with each other about what brought them to MIT, and what their hopes are for the upcoming school year. As the year progresses, other cohorts of students, staff, faculty, and alumni will come together for similar conversations as well as those initiated by the community on areas and topics of common interest.

The realtalk@MIT program is run by a small team at the MIT Center for Constructive Communication, in collaboration with people and organizations across the MIT campus. As a community-driven project, the direction of realtalk@MIT is informed by student, staff, faculty, administration, and alumni leaders. The program leverages a conversation platform developed with, and operated by, CCC’s closely affiliated non-profit Cortico.

The long-term goal of realtalk@MIT is to establish trusted communication channels and build a culture of listening and shared understanding across boundaries at MIT. The program is designed to give all members of our community the opportunity to hear – and be heard. Stories emerging from realtalk conversations are expected to reflect a wide range of life experiences, enabling participants who share their stories – and who listen to others’ – to discover new connections and hear new perspectives within the community. realtalkt@MIT also provides participants an opportunity to be heard – elevating stories that leaders may not otherwise hear, and building a constellation of MIT voices that reflect the full diversity of our experiences.

What sets realtalk@MIT apart from other dialogue programs is the unique combination of “ancient” methodologies (from facilitated dialogue, deep listening, and organizing) and digital technologies (including AI and data visualization). In fact, realtalk@MIT can be used to support and enhance the work of other dialogue programs that may benefit from recording, highlighting, and analyzing themes that arise from multiple conversations.

The realtalk@MIT program is for the entire MIT community. Students, staff, faculty, administrators, and alumni all bring unique and important experiences and perspectives to MIT’s culture. The program is open to any individual or group at MIT who would like to engage in small group conversations with members of the community.

realtalk@MIT will roll out during the fall term of 2024, when incoming first-year undergraduate and graduate students will be invited to participate in realtalk conversations, giving them a unique way to connect with their classmates beginning their very first week on campus. During the upcoming school year, realtalk@MIT will expand through additional community-initiated programs – e.g., students, staff, faculty, administrators, and possibly alumni. All new participants will be invited to respond to the realtalk “kickoff” questions – such as the values you bring to MIT and the experiences behind them. Over time, the program will broaden to include more specific conversations based on community needs and interests.  

realtalk@MIT has been designed to give you, and all members of our community, the opportunity to hear – and be heard. Stories emerging from realtalk conversations reflect a wide range of life experiences. By getting involved – sharing your story in conversation and listening to others’ – you will discover new connections and hear new perspectives among your peers and within the community at large. 

realtalk@MIT also provides participants an opportunity to be heard (via highlights recorded and shared with their consent). Getting involved means elevating stories that leaders may not otherwise hear, contributing your voice to a constellation of MIT voices brought to life in audio medleys and data visualizations – and ultimately helping to strengthen a culture of listening and shared understanding at MIT.  

The program will progress to offer students, staff, faculty, and alumni opportunities to dive deeper into realtalk’s methods and tools, develop facilitation and sensemaking skills, and bring those skills and tools into their own communities for conversation projects about issues important to them.

Go to the realtalk@MIT webpage or email realtalk@mit.edu to learn more about the program, join our mailing list, and express your interest in getting involved as a participant; getting trained as a conversation leader or as sensemaker; or bringing realtalk@MIT to your dorm, group, or department.

CCC’s guiding philosophy is that hearing the humanity in others through trustworthy communication channels is necessary for democracy, for communities, and for institutions to function. Informed by years of social media and media analytics, CCC started work in 2016 on designing new spaces that foster constructive communication and surface traditionally underheard voices into public dialogue and institutional decision making.

In spring 2022, CCC launched a realtalk@MIT pilot with the support of the MIT Values Statement Committee and the Office of the Provost. This pilot included more than 160 community members who participated in 50 small-group conversations led by 44 facilitators across the Sloan School of Management, the Department of Aeronautics and Astronautics, the Department of Urban Studies and Planning, the Media Lab, and the Undergraduate Association. The idea was to develop the realtalk system and test the use of conversations as a vehicle for listening to others and for improving decisions that impact the quality of life at MIT. Conversations prompted people to share personal stories that related to MIT values: What needs to be celebrated and what causes friction? What needs to change? How might we imagine the future of MIT?

This pilot helped lay the groundwork for establishing a culture of open communication and active listening, provided participants with a sense of empowerment, and established a trustworthy model for moving forward. Over the following two years, the Sloan School and the Media Lab have continued this work. The pilot program helped build trust and connections among participants, set the foundations for a scalable and adaptable framework for wider adoption, and garnered broader support from MIT  leadership. The current expansion of realtalk@MIT builds on this initial work.

The initial realtalk@MIT pilot was funded by the MIT Values Statement Committee and the Office of the Provost, with strong staff and student support from CCC. To expand realtalk’s deployment across MIT, the program received funding in 2024 from the Office of the President and the Office of the Provost to support additional personnel and operations.

No. We believe in the value of a pluralistic democracy, and that hearing the humanity in others is necessary for democracies to function. 

There are three steps in the realtalk@MIT process: 

1] Talk: Share your voice with a small group of peers, in person or virtual. 

Participants are organized into small-group conversations (~60-90 minutes, in person or on video, with recorded audio) among people with common bonds or experiences, such as first-year students, dormmates, affinity groups, or affiliate groups. The conversations are led by trained MIT community members who encourage sharing personal experiences over opinions, which research shows increases empathy, respect, and trust. 

2] Understand: Leverage human-led, technology-assisted methods to identify patterns and highlight key comments from across all conversations.

Post-conversation, participants actively review the recording and transcript to highlight key comments from across all conversations. This is always done with the consent of the speaker. Building on this foundation, trained realtalk “community sensemakers” use AI tools to find themes across these highlights, surfacing commonalities between people and their experiences that might otherwise go unnoticed.

3] Share: Hear audio medleys of voices from across the MIT community, and explore patterns and insights from a growing collection of conversation highlights.

Community members will be able to access output from community sensemaking – e.g., themes and representative highlights emerging from a growing collection of conversations – in a variety of formats, including thematic/topical audio medleys, online voice portals, interactive installations, and data visualizations.

Participants are organized into small-group conversations (~60-90 minutes, in person or on video, with recorded audio) among people with common bonds or experiences, such as first-year students, dormmates, or affiliate groups. Groups are typically 4-6 people although smaller and larger groups also work. The conversations are led by trained MIT community members who encourage sharing personal experiences and stories over opinions – which research shows increases empathy, respect, and trust. At the end of a conversation, participants actively review the recording and transcript for key highlights to share out to the wider MIT community with consent from the person who spoke.

Conversation Leaders facilitate the small-group conversations that are at the heart of realtalk. As the facilitator, the Conversation Leader guides the group through the process: helping everyone feel comfortable; establishing guidelines; asking questions and modeling responses; and supporting constructive communication. Conversation Leaders support the community, develop valuable facilitation skills, and build strong connections by hearing others’ stories. Anyone can become a Conversation Leader and it is an excellent opportunity to meet people from across the MIT campus.

Conversation participants consent to having their conversations recorded so that they may elevate one another’s stories for our community and our leaders to hear. Human-steered AI tools then surface patterns and themes across these recorded stories, while audio medleys, data visualizations and other outputs amplify voices for the community to hear.

For participants, audio recording enables you to extend your voice beyond a small-group conversation and make a broader impact in our community. Shared with your consent, a story you choose to tell can become:

• a connection you discover – for example, through an interactive realtalk highlights map – to others in the MIT community with similar stories or interests
• a new perspective that is heard and understood – for example, through a realtalk insights portal –  by others and by leaders at MIT

Recording also enable us as a community to listen more actively and effectively, capturing our collective thoughts to:

• surface insights (via audio medleys, data visualizations, interactive portals) that help us better understand each other across differences
• amplify often underheard voices as input for informed decision making and impact for change within the Institute

Once a conversation has taken place, the recording gets uploaded to a conversation platform developed and implemented by Cortico, a non-profit closely affiliated with CCC that is supporting realtalk@MIT. Access policies for conversation recordings and transcripts vary by project, and will be presented clearly and transparently for participant consent prior to a conversation.  

For the first-year undergraduate orientation project, for example, everyone who agrees to participate in a small-group conversation, plus the realtalk@MIT program team, is able to access the recordings and transcripts of all of these conversations. So if 50 first-year undergraduate student conversations occur during orientation, every participant will be able to hear audio and see transcripts of all 50 conversations.

Other realtalk@MIT conversation projects (e.g., among student, staff, faculty groups) may have different access policies – with the degree of data visibility and access varying according to the goals of the group and the project.

• First: You can choose to participate in realtalk, or not to participate at all.
• Second: If you choose to participate, you are consenting to participate in voice recordings / transcriptions of your small-group conversations 
• You will see a few other options that follow:  
  • You can opt into a research program – but to be clear, this is completely optional and independent of your participation in Option B
  • You can opt to have your small-group highlights shared beyond your small-group conversation to all first-year participants — this will be on a per-comment basis 
  • You can opt to have your small-group highlights shared to the whole MIT community — on a per-comment basis
  • You could be invited to share your highlights and comments publicly, in which case you will have the choice whether to consent to that additional level of sharing  

By default, only other participants in your small group – and a limited number of realtalk@MIT program facilitators and Fora app / Cortico staff, to help make the overall program and software work at its best. 

You will have the option, with your consent, to further share portions of your voice recordings and transcripts with: 

• All other participants in your cohort (e.g., all other first-years’ small-group conversations, or all participants in new grad students’ small-group conversations) – via a notice and consent feature in the app, on a comment-by-comment basis 
• The whole MIT community – via a notice and consent feature in the app, on a comment-by-comment basis  
• The general public – realtalk program managers might ask your permission 1-1 to share your contributions even beyond MIT itself (e.g. on MIT’s website, or in a podcast, or in other public outlets)

The process of recording, highlighting, and sharing with consent is meant to empower conversation participants to build relationships and strengthen trust. There are three types of consent involved in participating in a realtalk@MIT conversation – one consent that is necessary to participate in the program, and two others that are optional.

1] (Optional) Consent to participate and be recorded: Individuals first consent to participate in realtalk@MIT and have their small-group conversations recorded as part of a project – e.g., the first-year undergraduate orientation. This consent allows all other participants to hear audio and see transcripts of all recordings in that project, and also enables human-steered AI and data visualization tools to surface patterns and themes across their stories and reflections.

2] (Optional) Consent to allow collected data to be used by CCC for research purposes: Those who agree to participate can give their additional consent to allow speech recordings, transcripts, app interaction data, and demographic information generated from realtalk@MIT conversations to be utilized to understand interaction patterns, assess impact, and enhance future tools and methods of the realtalk@MIT program.

3] (Optional per highlight) Consent to share recorded highlights: Post-conversation, participants then review an AI-generated transcript of what has been shared, highlighting any excerpts that would be especially meaningful and valuable for others to hear. The platform will request a specific consent from the speaker of a given highlight to share more widely. If they give consent, the highlight will be accessible by anyone in the MIT community to see. If they do not consent, the highlight will not be shared more widely.

Participants may elect to use a pseudonym in lieu of their name when they set up their profile on the realtalk platform. It is important to keep in mind, however, that a person may be identified from their voice even if they choose to use a pseudonym. A voice anonymization option is expected to be added in 2025.

If a participant says something during a conversation they wish they hadn’t, the realtatlk@MIT program team will work with the participant to redact a portion of the conversation. Participants who would like to request a redaction should contact realtalk@mit.edu, specifying what they like to remove (to the best of their memory) and provide details on which conversation they were in.

The realtalk@MIT program leverages AI – specifically large language models – and data visualization tools to help human sensemakers derive meaning from recorded conversations at scale. Qualitative analysis of tens/hundreds/thousands of hours of recorded audio is incredibly complex and time-consuming. CCC, in collaboration with Cortico, have developed AI “power tools” that help sensemakers discover and interpret patterns and themes across conversations. The aim of this approach – which continues to be researched at CCC – is for AI support that allows sensemakers to be more efficient, confident, consistent, collaborative, and accurate with their analysis across large volumes of conversation data. 

CCC’s philosophy on AI generally, and for realtalk@MIT specifically, is for humans to remain in control at every step of the process, steering the technology based on their understanding of the context. Fully automated sensemaking is likely to miss critical community knowledge central to accurate and inclusive interpretation.  

Finally, the goal of AI is not just to make sensemaking faster, but also to make qualitative analysis techniques accessible with a manageable amount of training. A key part of the realtalk@MIT program is to enable community members – students, staff, faculty – to be trained and certified as sensemakers and participate actively in the analytical process. This approach to sensemaking signals a transition from conventional analysis methods to community-powered understanding.

Concerns about bad actors are understandable, and exist across most communications forums, ranging from social media, to large email/Slack distributions, to town hall-style meetings. 

CCC has taken various measures to guard against bad actors in realtalk, starting with its core design principles. The full realtalk@MIT framework – structured, facilitated, face-to-face, small-group conversations – has been designed to promote constructive dialogue. For participants who choose Option B – the tech-enabled version of realtalk@MIT that includes recordings, transcriptions, highlight, sensemaking, and the option to further share one’s voice to the MIT community – the three-stage consent process described above has been designed to reduce the risk of anyone’s voice or story spreading without their approval. But even with these designs in place, there are no guarantees against someone choosing to thwart the usage terms and violate the spirit of the realtalk@MIT program.  The realtalk@MIT team is committed to providing support to mitigate and ameliorate harm.

Meaningful sharing experiences typically involve risk. While some amount of risk is involved for participants, realtalk’s structure is designed to minimize risk, and is based on the belief that the benefits – greater connection and empowerment – ultimately outweigh the risks.

To be clear, a core reason to participate in a realtalk conversation is to have your voice shared; if you want to talk about something that is private or highly sensitive, you should seek other forums and formats that are designed for those purposes. 

As the conversation concludes, participants may ask to share demographic information to help put conversation insights in context and uplift diverse voices within the MIT community. CCC is committed to keeping all participant information private, with this information being stored separately from the conversation data and is only accessible by the realtalk@MIT team at CCC. Any demographic information shared as part of public-facing reports or public talks/events will be presented in aggregate.

Although the primary purpose of realtalk@MIT is not research, participants in the realtalk@MIT program will be given the opportunity to provide consent for allowing their data to be used by CCC for research purposes.  

Speech recordings, transcripts, app interaction data, and demographic data generated as part of realtalk@MIT will be used for the purposes of understanding interaction patterns, evaluating impact, and improving future tools and methods of the realtalk@MIT program. This includes analyzing communication styles, engagement levels, and the overall effectiveness of the program components. 

Your decision to include your data in this research is completely voluntary. If you choose to provide access to your data, you may subsequently withdraw your data from the study at any time without penalty or consequences of any kind. If you choose not to include your data for research, that will not affect your ability to participate in the realtalk@MIT program or any other aspect of your relationship with MIT. The CCC team has been developing a detailed research protocol with COUHES under the IRB # 2406001337 and the title: “realtalk@MIT: Transforming MIT into a Listening Campus.” Professor Deb Roy, who directs CCC, serves as the PI for this work.

No. Opting in includes recording and transcriptions, those Conversation Records will be used only to make the realtalk@MIT program as described here possible, and in line with your follow-on sharing consents, if any.

The realtalk@MIT team is prepared to flag sensitive information in two categories:

1] A participant in a conversation says something that could harm them if shared widely. When such content is flagged by anyone – a facilitator, sensemaker, realtalk team member, or even a member of the community upon hearing a highlight – the realtalk team will flag the concern with the speaker, make sure they want the highlight to be shared, and unshare and/or redact depending on the wishes of the speaker.

2] A person shares a reportable incident that an MIT office would typically address – e.g., a specific assertion of discriminatory or abusive behavior on campus or specific thoughts of self-harm. In such situations, the realtalk@MIT team will file a general report to the appropriate office, remove and delete the recorded content from the realtalk program, let the participants of the conversation know this has been done, and give the person who shared the reportable incident guidance on where at MIT (e.g., IDHR, HR, MIT Health) they should redirect their report.