Approx. 4 minute reading time

4 min

New technologies, such as artificial intelligence and machine learning, can make humanitarian work better, faster and more effective, but there are dangers too, which is why the ICRC is proceeding with caution. Various departments across the ICRC – among them Digital Transformation & Data, Protection, Information and Communication Technology, Law and Policy – are grappling with these innovations and their implications in different ways. Here we look at one example.

When the ICRC recently began testing artificial intelligence (AI) to help reunite lost loved-ones in the wake of the European migration crisis, it represented a breakthrough technological opportunity. The ICRC’s pioneering Trace the Face website was already helping people who were looking for their missing relatives to post their own photo online in hopes of being recognized and reconnected. Recently, the application of artificial intelligence for facial recognition to automate searching and matching is revealing further potential of new technology to accelerate one of the ICRC’s core activities.

“This was the starting point of our exploration,” says the ICRC’s Strategic Technology Advisor Vincent Graf Narbel. “What if we could improve the work of our field colleagues and provide a way to assist them by matching people’s faces using AI facial recognition techniques?”

Not long ago, separated family members had to sift through pages of photographs in physical books but that is changing, fast. The ICRC’s current database of people separated in the escape from war and hardship, part of which is accessible through Trace the Face website and touchscreen kiosks known as Trace the Face Corners, is helping to speed up this process and reunite more families than ever before.

“Trace the Face was the starting point,” says Graf Narbel. “The objective has shifted now to something much broader.”

Reuniting victims of war all over the world is among the ICRC’s most important functions, and remains even more so as the number of lost, disappeared, and missing continues to grow.

Informing these continued efforts is the evolution of projects such as AI for facial recognition that is revealing a range of new questions and challenges coexisting with the potential benefits of this brave new world.

Power and pitfalls

As seen in the trial for Trace the Face, AI-based facial recognition and name processing and matching technologies have obvious applications in filtering the database for duplicates and connections, so what’s the problem?

“Just because you can do something, it doesn’t mean you should,” says Graf Narbel. “There are data protection issues that are very specific to the ICRC.”

With Trace the Face, no photo can be published without the individual’s consent. But there is much personal data in the world that does not have consent, is not protected, and can be open to use – including abuse – by third parties.

This is a grave concern. “We have to be very cautious with our metadata and database security,” says Graf Narbel. “The question is how we can leverage these technologies while safeguarding the personal data of individuals – a critical aspect of protecting their rights – and adhering to International Humanitarian Law. We are exploring the power and pitfalls of technology.”

That exploration is simultaneously underway across other parts of the ICRC, including in relation to humanitarian law and policy, and armed conflict.

Do no digital harm

Working solely within its own database is one way the ICRC can ensure that protection standards are upheld, but this also means ignoring a plethora of other potential sources – such as Facebook or similar social media platforms – that could make finding the missing more successful.

The ICRC has always been concerned with protecting the people it serves from the physical dangers implicit in surviving conflicts, crises and disasters, but nowadays there is the new threat of virtual risk leading to actual physical harm.

In June, the ICRC published ‘Artificial intelligence and machine learning in armed conflict: A human-centered approach’, a report seeking to identify the potential risks as new technologies are applied in conflict scenarios.

“At a time of increasing conflict and rapid technological change, the ICRC needs both to understand the impact of new technologies on people affected by armed conflict and to design humanitarian solutions that address the needs of the most vulnerable,” the report states.

It is in this context of caution, thoughtfulness and a willingness to go slowly to get it right that the ICRC is seeking to situate its own application of artificial intelligence and machine learning. “There is a growing recognition of the power of technology to improve the work, but also a recognition of the dangers of technology to the people we’re serving,” says Graf Narbel.