In collaboration with AWS, FINCA is harnessing AI to conduct evidence-based customer research and creating new ways for nonprofits to address the world’s most pressing challenges.
Generative AI has the potential to help solve the world’s most pressing problems, but only if we think creatively and put the right resources behind it. FINCA, on a mission to eradicate global poverty, is teaming up with Amazon Web Services (AWS) to develop new AI models that help us get to the root of our customers’ challenges and lift their voices within the organization.
We are working with technical experts at AWS to bridge pressing gaps in our understanding of client-level impact through an Automated Research Assistant (AuRA). This AI-powered tool helps us gain deep knowledge of what our clients need and prove that our work is creating the best pathways out of poverty.
Earlier this month, FINCA Vice President of Research and Data Science Scott Graham, AWS Managing Director of Nonprofits and Global Healthcare Allyson Fryhoff, and AWS Senior Solutions Architect and Team Lead Han Tran joined the main stage at the 2024 AWS Re:invent conference to share AuRA’s approach of combining generative AI capabilities with subject matter expertise to map new pathways out of poverty and scale social impact.
Using Technology To Lift the Voice of Our Customers
How can generative AI help us understand and solve an intractable problem like poverty? For years, FINCA has been using technology to reduce the time it takes to do essential research and to assist with measuring and understanding the impact of our solutions.
Most of FINCA’s 9,000+ staff live and work in the communities they serve, and our research teams continuously survey our customers to understand their struggles and their successes. Yet, despite the thousands of FINCA boots on the ground, surveying clients living in some of the most rural places on Earth — who are often illiterate and lack connectivity or technology — presents unique difficulties. Add in many local languages, misinterpreted questions, and accidental wrong answers, and the survey data can get murky indeed.
To improve data quality, FINCA’s research team created ValiData — a platform for gathering and analyzing field data in difficult operating environments. But even with this significant advancement, other challenges came to the forefront — specifically a lack of a scalable, flexible infrastructure for data analysis. Looking to update our research methods, particularly with cloud computing, FINCA teamed up with AWS to guide us through a fundamental revamping of our data architecture — including automation and better connections to external sources of data — based on a new strategy for using technology to accelerate our knowledge.
In 2022, FINCA was selected as a winner of the AWS IMAGINE Grant for nonprofits that are using technology to solve the world’s most pressing problems. We seized the opportunity to collaborate further with AWS to build a cloud-based infrastructure for ValiData that now enables dynamic dashboards for tracking poverty, financial health, and women’s economic empowerment indicators, offering a comprehensive representation of the voices of our millions of clients.
Revolutionizing Research Through Cloud-Based Solutions and Generative AI
Propelled by these wins, FINCA and AWS have elevated our vision far beyond the original concept with an AI-powered Automated Research Assistant (AuRA). This innovative platform enhances our research with a new set of powerful generative AI services, drastically accelerating the identification of evidence-backed poverty solutions.
To complement the field data analyses coming in through ValiData, FINCA researchers also look for evidence to demonstrate that FINCA’s work is having a demonstrable impact. This type of research includes identifying relevant source documents from databases and search engines, going through them one by one, tagging, filing, and sharing them, and summarizing their findings — traditionally a slow and laborious, yet necessary, task.
But fortunately, this process can be accelerated using generative AI and process automation. FINCA’s AuRA platform uses generative AI to expand researchers’ original search queries into a full set of topics and subtopics organized into an outline. AuRA then employs process automation to fast-track the identification, tagging, and storage of relevant documents, using the AI-generated outline to power its search engines. This results in faster, less labor-intensive, and more evidence-backed impact pathways, or theories of change.
The tool also allows researchers to give feedback to AuRA and to improve its sourcing of evidence. This interaction between subject-matter experts and data machines ultimately produces clearer, more trusted theories of change.
Plus, unlike broad-based AI — which excels in automating knowledge but often lacks access to critical data from underserved communities — FINCA’s model is iteratively trained on emerging anti-poverty strategies, women’s issues, and data from other underrepresented segments of the global population. Purposeful, curated datasets like those from ValiData, combined with human expertise, help guide the model away from harmful biases and ensure its training includes voices that would otherwise be absent.
Joining Forces for Transformative Impact
FINCA’s collaboration with AWS on ValiData and AuRA reflects a shared commitment to harnessing technology to create pathways out of poverty. These two platforms innovatively bridge a gap in development research by leveraging FINCA’s direct customer interactions to generate valuable insights into their needs and struggles and using those insights to better represent customers’ voices and address their unique challenges.
AWS provides the essential tools to develop these robust technological solutions and FINCA brings invaluable expertise and deep knowledge, proving that when cutting-edge technology is paired with purposeful, targeted data and curated AI models, the results can drive meaningful and transformative impact.
Check out FINCA’s session on AuRA at AWS re:Invent 2024:
This article first appeared on finca.org