A Human-First, AI-Enabled Approach to Fighting Poverty

Jul 7, 2026

Article by Poverty Stoplight Innovation Week

In this session, Zahid Torres-Rahman, Co-founder and CEO of Business Fights Poverty, presented a practical reflection on how artificial intelligence can be used to strengthen social impact without displacing people from the center.

Drawing on the experience of Business Fights Poverty, a global community of 30,000 professionals, Zahid explained how AI can facilitate partnerships, activate accumulated knowledge, and improve program implementation. At the same time, he emphasized that trust, empathy, judgment, and lived experience remain fundamentally human capabilities.

DOING MORE WITH LESS WITHOUT REDUCING AMBITION

Zahid began by describing the current context of the social sector. Climate crises, conflicts, the consequences of the pandemic, and other challenges overlap and have a greater impact on people who are already experiencing vulnerability.

At the same time, many organizations face budget cuts, reduced access to funding, and political resistance. This creates growing pressure to do more with fewer resources.

However, Zahid proposed framing the question differently:

“How can we maintain our ambition in a context of crisis and constraints?”

From this perspective, AI can provide capabilities that previously required weeks or months of work. It can help identify partners, analyze large volumes of information, draft proposals, recognize patterns, and personalize services. Its value lies not only in reducing costs, but also in enabling organizations to expand their reach and make better use of their resources.

THREE AREAS WHERE AI CAN STRENGTHEN IMPACT

Business Fights Poverty organizes its work around three functions: connecting people, facilitating learning, and improving implementation. Zahid explained how AI can expand each of these areas.

Connecting and collaborating: AI can identify potential partners, funders, and experts within minutes. It can also help organize information, prepare messages, and build a narrative for a proposal. However, technology cannot create the relationships of trust needed for a partnership on its own. Aligning interests, managing tensions, and building shared commitments remain human responsibilities.
Turning knowledge into learning: Many organizations have years of reports, evaluations, research, and experience that remain scattered or are not used systematically. AI can analyze this knowledge, synthesize lessons, compare them across countries, and make them accessible in different languages. It can also strengthen institutional memory so that new programs are designed with an understanding of what has and has not worked in the past.
Improving implementation: AI can reduce administrative tasks, free up time for frontline staff, personalize support, and adapt content to different contexts. As an example, Zahid mentioned a project that organizes more than 200 documents on human rights and living wages to provide suppliers with guidance adapted to their country and language.

“If we do not use the knowledge we have, it becomes dead knowledge.”

Zahid explained that Business Fights Poverty has used AI to make two decades of accumulated content and experience accessible. In all these cases, AI can expand human capacity, but it cannot replace judgment, empathy, or an understanding of reality.

AI AS AN AMPLIFIER, NOT A REPLACEMENT

Zahid proposed imagining a continuum. At one end are tasks that AI can perform almost entirely; at the other are functions that should remain exclusively in human hands. Between the two lies the area with the greatest potential: AI used to expand human capabilities.

Technology can suggest a strategy, but people must interpret whether it is appropriate for the context, adapt it, and implement it. It can process information, but it cannot replace the experience of those who know the communities.

“AI can amplify knowledge, but trust is what turns that knowledge into action.”

This principle also applies to partnerships. AI can identify potential connections, but human relationships are what turn them into genuine collaborations.

THE SHAPE FRAMEWORK FOR PEOPLE-CENTERED AI

Zahid shared a framework developed by Business Fights Poverty to guide the responsible use of AI. The model, called SHAPE, brings together five elements.

Understand the System: Before incorporating AI, it is necessary to analyze the environment in which it will be used. There are significant differences in connectivity, digital literacy, access to devices, technological capacity, regulation, and data governance. AI does not enter a neutral system; it can strengthen positive aspects, but it can also deepen existing inequalities. The first question should be whether AI is genuinely appropriate for the context and what additional investments would be required to use it inclusively.
Amplify Human capabilities: Technology should strengthen people’s judgment, relationships, and capabilities. As an example, Zahid mentioned the use of AI by healthcare workers. The tool can extend their reach and provide information, but trust, empathy, and an understanding of the patient’s experience come from the professional. The central question is whether AI increases people’s decision-making power and dignity or, on the contrary, weakens them.
Be guided by Accountability: Success should not be measured by whether the newest model is being used, but by the concrete results achieved. In some cases, a small, specialized tool may be more appropriate than a complex system. Mechanisms must be in place to review, challenge, or correct a decision produced by AI. People must retain the ability to intervene and reverse incorrect or unfair outcomes.
Work through Partnerships: No organization can manage all the opportunities and risks of AI on its own. Solutions require the participation of communities, technical specialists, social organizations, businesses, governments, and people with direct experience of the problem.
Place Equity at the center: Communities should not be treated solely as sources of data. They should participate in defining the problem, designing the tool, governing the information, and distributing the benefits.

“Communities must help define the problem, the design, and the benefits.”

Participation strengthens the relevance, trust, and sustainability of solutions. It also enables people to retain decision-making power over their own data and over how the technology is used.

PRACTICAL APPLICATIONS SHARED DURING THE SESSION

Zahid presented several examples of tools that combine AI and human collaboration.

Business Fights Poverty is using AI to identify potential funders and partners for projects in Africa. The technology helps organize information and prepare messages, while trusted networks make it possible to bring people together and build agreements.

It is also working with a financial organization to turn ten years of experience supporting entrepreneurs into a learning tool available across different countries and languages.

Another project seeks to use AI to expand the application of the Poverty Stoplight, known as Compass within the Business Fights Poverty network. The proposal would make it possible to better understand people’s needs, direct resources, strengthen livelihoods, and analyze which interventions are working.

OPPORTUNITIES FOR THE POVERTY STOPLIGHT

Julia Corvalán, Global Operations Manager of the Poverty Stoplight, explained that Fundación Paraguaya considers AI an important tool for expanding the reach of the Poverty Stoplight.

The organization aims to support a global population far beyond the current capacity of its team. AI could therefore help it reach more families in a timely manner and at a more sustainable cost.

One possibility is to develop AI mentors or agents that provide initial support to facilitators and families. These tools could answer questions, provide guidance on using the platform, and facilitate access to information.

Fundación Paraguaya is already working with Rosa, an AI agent trained with information from the Poverty Stoplight. The tool is being improved with support from the team and partner organizations and is expected to become available to platform users.

Julia clarified that Rosa can make mistakes and that its answers must be reviewed. Having a large database does not automatically mean that useful learning will be generated.

“AI can provide information, but human knowledge and experience are needed to turn it into learning.”

The aspiration is to use AI as an enabler to reach more families without replacing human relationships or the development of individual agency.

CONDITIONS FOR USING AI RESPONSIBLY

The conversation identified several essential conditions:

Assess whether AI is appropriate for the context and population.
Keep people involved in decision-making.
Protect privacy, dignity, and autonomy.
Review the results generated by technology.
Recognize inequalities in connectivity and digital literacy.
Design with communities, not only for them.
Measure success according to real outcomes rather than the sophistication of the tool.
Prevent AI from replacing human relationships based on trust and empathy.

EXCHANGE WITH PARTICIPANTS

Instead of a traditional question-and-answer session, participants joined discussion groups in Spanish and English. They reflected on how artificial intelligence can help organizations maintain their ambition in a context of crisis, limited budgets, and growing social needs.

One of the main contributions identified was AI’s potential to analyze large volumes of information more quickly. In the case of the Poverty Stoplight, this could help transform recent family data into useful insights for organizations and the public sector, reducing the time needed to identify patterns, design responses, and adjust programs.

Participants also shared experiences of collaboration between people and AI. One participant explained how they used these tools to prepare a report on financial inclusion, enabling them to conduct more research and deepen the analysis within a limited timeframe. However, they noted that the results had to be continuously reviewed and corrected in order to turn the information generated into genuinely relevant conclusions.

Other participants highlighted the usefulness of AI for interpreting health and community program data, identifying potential partners, and quickly connecting with specialists. At the same time, the groups agreed that efficiency should not displace human judgment. Technology can expand capabilities, but people must continue to interpret the results, understand the context, and ensure that decisions respond to the real needs of communities.

Get Involved

Interested in attending future virtual and in-person events from the Global Stoplight Network? Subscribe to our monthly newsletter here.

You can learn more about the upcoming Cerrito Forum 2026, the annual international global development event that brings together global stakeholders committed to the elimination of poverty from October 12th - 16th by visiting cerrito.com.py.

Back to all webinars