Artificial Humanitarianism—The Data-Driven Future of Refugee Responses

In Jordan’s Za’atari refugee camp, to pay for food at one of the World Food Programme (WFP)’s distribution centers, Syrian refugees must lean into an iris scanner that registers their biometrics.

The scanner then logs their transaction onto the program’s blockchain network: Building Blocks.

Za’atari is the site of the world’s first comprehensive biometrics system for refugees. Unveiled in 2013 by the United Nations High Commissioner for Refugees (UNHCR), it included the large-scale digitization of refugee registration through the Biometric Identity Management System (BIMS). The UNHCR started using BIMS after a nearly 100-person trial of Afghan refugees in Pakistan in 2002.

The system works by creating a unique biometric identifier for each registered refugee. That identifier is then stored in a blockchain called the Population Registration and Identity Management Ecosystem (PRIMES). Initially pitched by the UNHCR as a way to easily track refugee registration and residency, the biometrics model has morphed into an all-encompassing surveillance system. BIMS registers every refugee life event: from marriage, to education, to death. More than 90 percent of Syrians residing in Za’tari have been forcibly registered into the system as a precondition for receiving aid.

As a laboratory for large-scale digital registration, the Za’atari model has now become the norm in UNHCR operations. Registration in BIMS is a prerequisite for refugee registration and thus the ability to apply for asylum or resettlement as well as for the receipt of humanitarian aid. By 2020, more than 37 million refugees around the world had given their biometric information and were registered through BIMS into PRIMES. Many of these refugees now receive aid payments with digital wallets, communicate to the UNHCR through chatbots and, in some rare cases, receive aid from drones.

Robotics, biometrics, drones, artificial intelligence, computational software and blockchain are not usually associated with humanitarianism but that is changing rapidly. Today, the humanitarian sector is imbued with a digital logic that is reshaping the delivery of aid and the conception of care. The overlapping catastrophes in the Middle East and North Africa—from the Syrian civil war of the past decade to the ongoing mass violence in Sudan and Palestine—are a key laboratory for testing these new practices.

As refugees are transformed into digital subjects, various technologies mediate their interactions with humanitarian organizations. The explicit goals of these intermediation technologies are to enhance the efficiency of humanitarian care, but they are also shaping the predictive, autonomous capabilities of technologies to meet humanitarian needs. To do so, they need a specific fuel: data.

 

‘Data is the New Oil’

 

Many technology observers have latched on to the expression that “data is the new oil” to reflect the growing importance of data-driven digital technologies in the economies of the Middle East.[1] The region’s commercial appetite for data extends into the humanitarian sector. Local companies have a hand in developing everything from mobile payment applications to biometric registration systems—like the Zayn wallets used for digital cash transfers, developed by an Iraqi company, or the EyePay systems used to collect iris and fingerprint data in Za’atari, developed by a Jordanian-British company. An entire network of commercial interests now exists to feed into the humanitarian sector’s desire to digitize everything from refugee registration to aid delivery.

States are getting in on the action too. The United Arab Emirates has become a hub for networks of government bodies, commercial interests, international consultants and local start-ups that deepen the overlapping circuits of war, surveillance, logistics, humanitarianism and digitization. For example, the International Humanitarian City in Dubai is a logistics project created in 2003 to connect aid to conflict and disaster zones in the Horn of Africa, Red Sea and the Indian Ocean region—part of a broader effort to expand influence and logistics infrastructure in Africa and Asia. Recently, the UAE created a parallel “digital response platform” that aims to provide a digital corollary to these logistics capacities. This platform brings together machine learning, artificial intelligence and geospatial technologies to facilitate information sharing among Humanitarian City’s partners.

The involvement of tech companies and their products in state-led and international humanitarian work raises serious concerns about how their profit motive might compromise humanitarian principles. In particular, the absence of strong data protection laws renders data a profitable commodity to be bought and sold. Any commercial company holding a contract with the UNHCR or WFP can access anonymized refugee data. The more data available to companies, the wider range of possible technological applications can be adapted to humanitarian use. But this data is rarely protected and can be circulated among humanitarian and commercial actors long after an individual is no longer receiving humanitarian support. Humanitarian emergencies thus become opportunities for companies to profit as both service providers and data hoarders.

The adoption of digital technologies also changes the nature of the humanitarian problems that actors are trying to address. One example is predictive analytics: a type of artificial intelligence in which the timing, cause and duration of anything—from an environmental catastrophe to a violent conflict— can be anticipated and prepared for. Using predictive analytics, the WFP has created a modelling platform called SHAPES (Shock and Assistance Platform for Economic Simulations) to forecast food security outcomes in the event of severe crisis. It has already deployed SHAPES in both Yemen and Lebanon, reducing the average food security assessment time from six months to three. It also uses the “Hunger Map Live” to model caloric needs depending on the location of the disaster. The United Nations Office for the Coordination of Humanitarian Affairs created the Humanitarian Data Exchange (HDE) to provide data and datasets to organizations like the WFP that feed into their predictive analytics. These programs all cast data and associated problems differently but share a commitment to forecasting how to deal with humanitarian emergencies.

In this incarnation of contemporary humanitarianism, each emergency produces data that feeds into the responses of humanitarian organizations anticipating the next emergency. The data is not only circulated through PRIMES, SHAPES, HDE and other similar platforms. It also becomes the source feeding predictive humanitarianism. As a result, knowledge about emergencies and those affected by them is no longer being produced in context but rather through machines and artificial intelligence that dictate how humanitarians should intervene in crisis situations. By producing predictive visions that can be acted upon by humanitarian organizations, these technologies shift the whole terrain of humanitarian action from reactive to anticipatory.  The cascading crises in the Middle East and North Africa provide a veritable data mine for humanitarian organizations to hone these predictive analytics.

Anticipating a crisis and acting upon one are very different things, however. Predictive analytics will always rely on incomplete data. They will also include algorithmic biases, which have already been called into question in the case of predictive policing that overwhelmingly targets racialized and poor communities for police interventions. Furthermore, the data fed into these predictive algorithms has often been extracted from people in humanitarian emergencies that may have not understood or consented to the future uses of their data.

 

Humanitarian Robots

 

In 2013, the US Defense Department’s Defense Advanced Research Projects Agency (DARPA) created a Robotics Challenge to incentivize commercial and academic research into the potential for these technologies to support humanitarian and disaster relief operations. Teams submitted everything from robotic snakes capable of capturing images under rubble, to humanoid robots that could drive cars, use power tools and transform into moving vehicles. The robots might have been clunky, their movements slow and awkward, but they showcased a new way of being humanitarian.

Many of DARPA’s publicly known projects are the stuff of dystopian science fiction, but the robotics envisioned in the 2013 challenge were not from the realm of fantasy or techno-futurism. A robot named Sophia is the first United Nations Development Programme’s (UNDP) Innovation Ambassador. Notably, she is also the first robot to receive legal personhood, after Saudi Arabia granted her citizenship in 2017. Sophia mingles with dignitaries and diplomats at various UNDP events. At an “AI for Good” summit in July of 2023, Sophia was joined by eight other humanitarian robot prototypes that can do everything from prepare food, heal wounds or simply socialize with someone who needs a friend. A future in which robots are an integrated part of the landscape of refugee camps, serving as soccer players, companion dogs or cleaners, may not be far off.

But humanitarian robots can do much more than socialize. Robotic technologies are justified as being able to move aid the so-called last mile, usually conceived as up to 50 kilometers, when it is too dangerous for humans to physically move aid into war or disaster zones. According to the head of the WFP’s innovation department, Bernhard Kowatsch, the idea for robotic vehicles as aid deliverers was conceived during the brutal four-year battle of Aleppo between 2012–2016, when it was virtually impossible to move aid into the city. A decade later, Project A.H.E.A.D (Autonomous Humanitarian Emergency Aid Devices) was born out of the WFP’s Business Accelerator and launched in Germany for its first use case in South Sudan. This pilot project involves remotely operating aid delivery into South Sudan using a Ukrainian built amphibious truck with an unassumingly playful name: Herbert the Sherp.

Herbert is one cog in a material structure linking several digital technologies and applications: Drones and satellites produce information about the terrain, that data is then transmitted to an operating center far from the afflicted area to determine what aid and supplies Herbert needs before the Sherp is loaded up and sent to the zone through remote operation. Humanitarian workers, by which I mean humans, are only one part of a mostly non-human, networked structure in which drones, satellites, communications towers and Herbert the Sherp are collecting the information needed to deliver humanitarian aid.

There are good reasons to believe that the Herberts of the humanitarian world will save lives and protect humanitarian workers in dangerous situations. But they also represent a new form of robotic intermediation that threatens to render life-and-death decisions, like who gets food and who does not, up to machine learning. Although humanitarian dilemmas about how to deliver aid and to whom have always existed, humanitarian robots change these calculations and potentially absolve humanitarian actors of the ethical considerations required to meet people’s needs in moments of emergency.

Moreover, these technologies are mirror images of those deployed in contemporary warfare—responsible for creating some of today’s most devastating humanitarian crises. Indeed, Israel has used remotely operated drones to surveil and target aid workers in Gaza and obstruct the delivery of aid. Cyber warfare targeting key infrastructures has become routine. Suicide drones are regularly deployed by Israel, Hizballah and other actors, and there is increasing evidence of underwater drones and land robots deployed on the battlefield.

The same technologies that network data, algorithms, machines and drones to recast humanitarian possibilities are thus similarly reshaping warfare. Israel’s “Lavender” and “the Gospel” are the most spectacular examples. Both systems are AI and machine learning programs that have identified more than 37,000 individual Palestinian targets that were subsequently targeted by Israeli drones. These systems function mostly independently of human oversight. Much like Project A.H.E.A.D, Lavender and the Gospel bring together data and machines to make decisions about life and death.

 

Circuits of Humanitarianism and Warfare

 

Za’atari demonstrated that large-scale digitization of registration and surveillance of everyday life by the UNHCR was possible. Its success meant that the model has been transported to other protracted refugee situations, including the Dadaab camp in Kenya, the Osire camp in Namibia and the Tongogara refugee camp in Zimbabwe.

Alongside the circulation of these technologies and practices are the ideas at the core of contemporary humanitarianism: accountability, transparency and efficiency. Humanitarian technophilia is rationalized as facilitating better forms of intermediation to meet people’s needs. But these rationalizations mask a complex system of power oriented toward data extraction and commodification. The immediate and long-term effects of crisis and emergency are rendered into a series of digitally legible data points that feed into systems of computational analytics.

Moreover, many of the same companies at the forefront of digital humanitarianism are implicated in digitized warfare and other forms of surveillance and violence. Palantir, for example, a company whose original funding came partly from CIA-linked venture capital firm Q-Tel, has contracts with such diverse customers as the Los Angeles Police Department and Immigration and Customs Enforcement in the United States to provide data analytics. Palantir is also a major WFP partner. Clearview AI is another company that provides surveillance technologies to police departments around the world and also provided facial recognition services to the Ukrainian government in the aftermath of the Russian invasion. The Israeli company Cellebrite exports border technologies around the world while also receiving contracts to service United Nations agencies with their software. Examples, such as these, that blur the lines between warfare, surveillance and humanitarianism abound.

Such overlaps aggravate concerns over how digital companies are recasting humanitarianism norms and principles. The large-scale destruction in Gaza, Lebanon and Sudan poses generational challenges for humanitarian actors who will certainly turn to digital humanitarian tools for solutions. For better and worse, each new crisis provides new opportunities for the deployment and refinement of these technologies.

 

[Samer Abboud is associate professor of global interdisciplinary studies at Villanova University.]

 

---

Endnotes

[1] Sisson, Patrick. “Desert Data Boom.” Sherwood News, June 27, 2024. https://sherwood.news/business/middle-east-data-centers-infrastructure-spending-demand/