AI for Social Impact

Understanding Human Experience Through Technology

My work in AI for social impact began with a simple observation: many of the most important aspects of human experience are difficult to capture through traditional data systems.

Communities often share rich stories about trust, aspiration, uncertainty, exclusion, and decision-making. Yet these insights frequently remain fragmented across surveys, reports, and videos, making them difficult to systematically learn from and incorporate into policy and program design.

This realization led me to explore a broader question: How can AI help organizations better understand the lived experiences behind the data?

Over time, my efforts evolved from an interest in storytelling and community voice to a deeper exploration of qualitative data, human-centered technology, and the role AI can play in helping organizations capture, analyze, and learn from complex human experiences.

Rather than viewing AI primarily as a tool for automation, I am interested in how it can help surface overlooked realities, strengthen understanding between institutions and communities, and support more informed and inclusive decision-making.

The goal is not simply to collect more information, but to better understand the people and contexts behind it.

Key Collaborations & Learning Partners

This work has been shaped through engagement with a diverse network of practitioners, researchers, technologists, journalists, and social impact organizations.

Notable collaborations and learning experiences include:

  • Video Volunteers — Community journalism, community voice, and participatory storytelling

  • 10x Impact Labs — AI for Social Impact Fellowship

  • Fields of View — Participatory systems thinking and governance

  • IDInsight — Data collection, evidence generation, and decision-making

  • Ooloi Labs — Technology and development innovation

  • Researchers, field enumerators, nonprofit leaders, and grassroots practitioners across India

These engagements have provided valuable insight into how information is collected, interpreted, and used across development ecosystems.

Research Conversations & Field Perspectives

A significant portion of research work has been informed by conversations with technology experts, field practitioners, and social impact leaders working at the intersection of technology and human development.

Jessica Mayberry, founder of Video Volunteers, has dedicated her work to strengthening community-based journalism and ensuring that the voices of marginalized groups are heard. Engaging with Video Volunteers has been among my most impactful and rewarding experiences, and Jessica’s mentorship has been instrumental in shaping my thinking on community voice, storytelling, and ethical engagement.

Vaibhav Mishra, co-founder of 10x Impact Labs, leads work at the intersection of A.I., data systems, and social impact. Through 10x Impact Labs’ A.I. for Impact Fellowship, I developed a strong foundation in systems-oriented approaches to designing A.I.-enabled data collection tools for understanding the lived experiences of rural and underserved communities. Vaibhav’s guidance and mentorship have been instrumental in shaping this work.

During a discussion I hosted in Bangalore on the role of technology and A.I. for rural data collection, Jahnavi Meher (Product Manager, IDInsight) reflects on how gendered power dynamics within households often limit women’s ability to participate in data collection processes. This insight mirrors broader field observations that structural and social constraints, rather than technical access, frequently determine whose voices are heard.

Kartik Natarajan (Co-Lead, Fields of View) highlights how data collection practices are inherently shaped by power and intent, and why communities may resist engagement when these dynamics are overlooked. His reflections underscore why responsible A.I.-enabled data collection must prioritize human judgment, ethical design, and trust over scale or automation.

Key Research Insights

Across interviews, field observations, events, and conversations, several themes have consistently emerged.

Participation is Social Before it is Technological

- Access to smartphones or internet connectivity does not automatically translate into meaningful participation

- Household dynamics, gender norms, trust, time constraints, and social context often play a greater role in determining whether people engage

Data Collection is Relational

- The quality of information is often shaped by the relationship between communities and those collecting information

- Field enumerators, community journalists, and facilitators play a critical role in building trust and creating conditions for meaningful participation

- Human interaction remains essential

Qualitative Insight is Difficult to Scale

- Many organizations collect rich qualitative information through interviews, surveys, field notes, videos, and community engagement

- However, synthesizing, organizing, and learning from these datasets remains challenging

- As a result, valuable knowledge often remains fragmented across teams, organizations, and projects.

Trust Shapes Data Quality

- Communities are more likely to participate when they understand how information will be used and when they see tangible value from sharing their experiences

- When data collection feels extractive or disconnected from local realities, participation often declines

AI Adoption Remains Challenging

- Many organizations recognize the potential of AI but face barriers related to capacity, trust, implementation, workflow integration, and organizational readiness

- The challenge is often not technological capability, but meaningful adoption

Understanding is Often More Important Than Collection

- Many institutions have access to more information than ever before

- The challenge increasingly lies in transforming that information into actionable understanding

Evolution of Thinking

This work has evolved significantly over time as new conversations, collaborations, and field observations challenged earlier assumptions about how AI might create value in social impact settings.

Phase 1: AI-Enabled Surveys & Community Data Collection

Through the AI for Social Impact Fellowship at 10x Impact Labs, initial exploration focused on how AI-powered chatbots and conversational interfaces could support community data collection through messaging platforms such as WhatsApp. The goal was to understand whether AI could help organizations collect information more efficiently from underserved communities while reducing the operational burden associated with traditional survey methods.

This raised broader questions:

- How should AI interact with communities?

- What barriers limit participation?

- How can qualitative information be collected responsibly?

- Does greater access to technology necessarily lead to greater engagement?

Phase 2: Supporting Human Data Collectors

As research progressed, it became increasingly clear that technology alone could not solve many of the challenges associated with community engagement and data collection. Field enumerators, community journalists, and local facilitators often play a critical role in building trust, understanding context, and encouraging meaningful participation. This shifted the focus away from replacing human interaction and toward exploring how AI might augment the work of those already operating closest to communities.

The question became:

- How can AI help people do their jobs better rather than remove them from the process altogether?

Phase 3: Video, Narratives & Qualitative Insight

Subsequent exploration examined how interviews, documentary footage, surveys, and field narratives could be treated not only as communication products but also as sources of qualitative insight. Many of the most valuable observations about trust, decision-making, aspirations, and lived experience often emerged through stories and conversations rather than structured survey responses

This led to growing interest in:

- Video as data

- Behavioral patterns

- Community-generated knowledge

- Qualitative insight extraction

Phase 4: Multi-Source Knowledge & Shared Learning

As engagement with organizations across the social sector expanded, another challenge became increasingly apparent.Many organizations were collecting valuable information through surveys, interviews, reports, videos, field observations, and community engagement efforts. Yet much of this knowledge remained fragmented across institutions and difficult to access or learn from collectively. This prompted a shift away from thinking solely about collecting new data and toward understanding how existing knowledge could be better organized, connected, and utilized.

The question evolved from how to collect more information to how do we learn more effectively from the information that already exists.

Phase 5: Community-Generated Evidence & Ground Truth

While existing datasets can provide valuable insights, they are often shaped by institutional priorities, methodologies, and assumptions. My documentary work increasingly highlighted the importance of generating original qualitative data directly through conversations, interviews, and community engagement. This phase focused on understanding how firsthand narratives and community-generated evidence can complement larger datasets and provide additional context around how people experience social, economic, and development issues. The objective became not simply to aggregate information, but to strengthen the connection between data and lived experience.

Phase 6: Trust, Verification & Shared Understanding

Current exploration focuses on how AI might help organizations aggregate, structure, verify, and learn from information collected across multiple sources while maintaining transparency, context, and trust. Many of the challenges encountered throughout this journey ultimately point toward a common issue: understanding. Communities, practitioners, researchers, nonprofits, and institutions often possess valuable but incomplete perspectives on the same issue. Rather than pursuing a single technology solution, this work has increasingly become an exploration of how AI can help connect these perspectives, transform fragmented observations into shared understanding, and support more informed decision-making.

Current Areas of Exploration

Narrative Intelligence

- How can AI help organizations identify patterns, themes, and emerging issues across interviews, stories, surveys, and field observations?

Video as Data

- How might documentary footage and recorded conversations become sources of qualitative insight beyond communication and storytelling?

Human-AI Collaboration

- How can AI augment the work of field researchers, community journalists, and practitioners while preserving human judgment and contextual understanding?

Community Knowledge Systems

- How can organizations better capture, organize, and learn from community-generated knowledge while respecting local ownership and participation?

Trust & Verification

- How might AI support greater transparency, accountability, and confidence in qualitative information collected across diverse stakeholder groups?