Are machine learning technologies ready to be used for humanitarian work and development? - Answering key barriers and risks

Aug 4, 2026

Upload by Aria

Novel digital data sources and tools such as machine learning and artificial intelligence have the potential to transform development data and improve humanitarian monitoring and response. The work surveys current state-of-the-art approaches and highlights barriers that must be addressed for data-driven methods to function reliably with imperfect, highly complex real-world data. It argues that without organized efforts, these technologies may underperform on promised goals and can worsen inequality, strengthen discrimination, and infringe human rights. The focus includes data and dataset challenges, definitions, and emerging sources like mobile and satellite data.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages15

Reading Time38 min

AccessFree to download

Are machine learning technologies ready to be used for humanitarian work and development?

Potential and risks of data-driven AI

New tools and datasets

Data challenges and emerging digital sources

Privacy, inequality, and real-world barriers

arXiv :2307 .01891v1 [physics .soc-ph] 4 Jul 2023 Are machine learning technologies ready to be used for humanitarian work and development? Vedran Sekara 1 ;2 ;􀀃 , Mrton Karsai3 ;4 , Esteban Moro5 ;6 , Dohyung Kim 1 , Enrique Delamonica 1 , Manuel Cebrian7 , Miguel Luengo-Oroz8 , Rebeca Moreno Jimnez9 & Manuel Garcia-Herranz 1 ;􀀃 1 UNICEF, New York, USA 2IT University of Copenhagen, Denmark 3 Central European University, Vienna, Austria 4Rnyi Institute of Mathematics, Budapest, Austria 5 Connection Science, Massachusetts Institute of Technology, Cambridge, MA, USA 6Department of Mathematics & GISC, Universidad Carlos III de Madrid, Spain 7Department of Statistics, Universidad Carlos III de Madrid, Spain 8 United Nations Global Pulse, New York, USA 9 UNHCR, Geneva, Switzerland *Correspondence should be addressed to these authors Novel digital data sources and tools like machine learning (ML) and artiﬁcial intelligence (AI) have the potential to revolutionize data about development and can contribute to monitoring and mitigating humanitarian problems. The potential of applying novel technologies to solving some of humanity's most pressing issues has garnered interest outside the traditional disciplines studying and working on international development. Today, scientiﬁc communities in ﬁelds like Computational Social Science, Network Science, Complex Systems, Human Computer Interaction, Machine Learning, and the broader AI ﬁeld are increasingly starting to pay attention to these pressing issues. However, are sophisticated data driven tools ready to be used for solving real-world problems with imperfect data and of staggering complexity? We outline the current state-of-the-art and identify barriers, which need to be surmounted in order for data-driven technologies to become useful in humanitarian and development contexts. We argue that, without organized and purposeful efforts, these new technologies risk at best falling short of promised goals, at worst they can increase inequality, amplify discrimination, and infringe upon human rights. New Tools and Datasets Data is critical for humanitarian and development work. Accurate and updated estimates of population demographics are vital in order to understand and respond to social and economic inequalities 1 , and to move from reactive to proactive interventions that mitigate the impact of crises before they happen2. As such, whether using global estimates of poverty to advocate for efforts or design policies to eliminate it, or using malnutrition data in the midst of a conﬂict to allocate resources to where they are most needed, data is at the core of the organizations that work to meet the 2030 Agenda for Sustainable Development3. It can, however, be hard to obtain accurate and timely data. In many parts of the world traditional household surveys are the main, and often only, method for demographic data collection. Surveys provide rich and irreplaceable data, but they can be expensive and time-consuming. As such, there is a growing focus on leveraging different big digital datasets and new tools like AI and ML to complement household surveys. This is particularity important in rapidly changing contexts (e.g. humanitarian crises or pandemics) as data and information can be retrieved and analyzed in fast and relatively inexpensive ways. Unfortunately there are no clear deﬁnitions of AI. In general terms they refer to systems, which sift through data, recognize patterns, and possibly make decisions based on their discoveries. This covers the full spectrum of models, from relatively simple statistical models (e.g. linear regression and decision trees) to more sophisticated, but still explainable mathematical models, to black-box like neural network and deep learning approaches. Technologies, like mobile phones, are starting to have signiﬁcant global coverage. Today there are 107 mobile-cellular subscriptions per 100 inhabitants worldwide4 (see Fig. 1A)

FAQs

What opportunity do machine learning and AI provide for humanitarian and development work?

They can revolutionize development and humanitarian data by enabling better monitoring and mitigation of crises, using novel digital tools and datasets.

What main barriers must be overcome for these technologies to work in practice?

Tools must become usable with imperfect, highly complex real-world data, and the report identifies barriers that currently limit reliable application.

What risks does the report warn about if these technologies are adopted without coordinated efforts?

Without organized and purposeful efforts, the technologies may fall short of goals and can increase inequality, amplify discrimination, and infringe human rights.

Aug 4, 2026

Upload by Aria