Innovating with Everyday Data: A New Frontier for Targeting Humanitarian Aid

One of the recurring themes in HOISA's discussions on the use of emerging technology and artificial intelligence in the humanitarian sector has been the lack of complete and unbiased data sets. Regardless of the tool used, its effectiveness depends on the quality and availability of the information that supports it. This problem is more visible in humanitarian settings, where data is often incomplete, outdated, missing, or extremely hard to gather. In wake of further funding cuts in the development sector, data collection on the ground is becoming even more difficult. HOISA’s recent panel discussion on “Innovating with Everyday Data: A New Frontier for Targeting Humanitarian Aid” attempted to engage with this shortcoming.

The conversation featured Professor Ahmed Mushfiq Mobarak, Professor and Economist from Yale University, Dr. Dibyadyuti Roy – Associate Professor and Critical AI Scholar from University of Leeds, and Mihir Bhatt – Founder and Director of AIDMI. They explored how the sector could innovate to use everyday data for humanitarian aid for better aid targeting. This question finds its roots in the impactful research and work that Professor Mobarak carried out in Bangladesh.

For several years, governments and aid agencies have relied on household surveys, community-based targeting and proxy means testing to identify vulnerable populations. While these methods are effective, depending on the scale and region, these can also be very expensive and slow. Mobarak's work investigates whether everyday digital footprints, such as mobile phone data, can help bridge some of these information gaps.

Everyday activities like phone top-ups, call patterns, and network usage provide valuable insights into a household's economic situation. A poorer household would have a very different way of using their phones, compared to a middle-class or rich family. For example, “Poor people tend to top up their phone more frequently in smaller increments due to cash constraints, whereas richer people might top up more infrequently but in larger increments, because it’s inconvenient to keep topping up your phone.” By analyzing these patterns, researchers managed to enhance the targeting of assistance in areas where traditional data collection would have been challenging or too costly.

What makes this work especially relevant for the humanitarian sector is that it brings forth a new way of looking at data. In resource-limited environments, such creative uses of everyday digital footprint of the human population can provide a lot of insights that the humanitarian sector can use.

The case study in Bangladesh reflects how critical collaboration among researchers, engineers, governments, and humanitarian organizations is to allow for such projects. Accessing data, navigating legal barriers, building trust, and turning findings into operational decisions all relies on teamwork across various sectors. This work stands as an important case study for humanitarian workers to think upon.

But while this opens up new frontiers for targeting humanitarian aid in regions such as South Asia, Dr. Dibyadyuti Roy in his presentation helped demystify the hype around the potentials of AI finding quick and all solutions for us. Tracing the history of AI, Roy points out that the technologies we are witnessing today rely heavily on human effort, labour and exploitation. Referring to the Ghost workers labeling datasets, reviewing content, and training models, Roy argues that this labour is often left out in important discussions. Further concerns about privacy and data protection also surfaced during the conversation. If humanitarian organizations increasingly depend on digital footprints to make decisions, what protections are necessary to ensure vulnerable populations are safe? Who owns the data? Who benefits from its use? And who faces the risks when issues arise?

As found in the study by Prof Mobarak, AI systems are also very context specific. A model that performs well in one country may behave differently elsewhere. The varying results from Bangladesh and Togo discussed during the session provided a clear example of this reality. Therefore, since AI systems cannot simply be transferred from one context to other, the enormous costs associated with setting up, training and maintaining the AI systems need to be considered while making decisions.

Mihir Bhatt connected these ideas by stressing that the true goal of innovation in humanitarian work is not technological progress but human well-being. Better targeting is important because it can help limited resources reach those who need them most. However, efficiency should not be the sole measure of success. Fairness, accountability, and justice must remain central concerns.

The discussion left participants feeling both hopeful and accountable. Everyday data presents real opportunities to enhance humanitarian decision-making, especially in places where traditional information systems are weak. Yet, the conversation also highlighted that technology is just one part of the equation. How these tools are designed, managed, and utilized will determine whether they strengthen humanitarian action or create new forms of exclusion. For humanitarian organizations facing an increasingly digital future, this may be the most crucial lesson of all.

Contributions from Mihir Bhatt, Ahmed Mushfiq Mobarak, Dr. Dibyadyuti Roy and Khayal Trivedi