How AI Is Making Humanitarian Response More Effective at Saving Lives
A helicopter coming in to land on scrubland, dust whirling up into the air. Hundreds of people on the move in the background; the elderly, young, and those supporting the sick. These are the images that confront us on an all too frequent basis as natural disasters ravage varied regions around the world. It may not be able to stop them occurring, but could AI help relief agencies to work more effectively and save more lives?
In 2017, there were around thirteen storms that raged the Caribbean alone, leaving millions of people without food, water or shelter. Disaster relief groups rely on accurate data to ensure that shelter, food and medical supplies are provided in sufficient quantities to support the populous area. As infrastructure collapses, people often move to new locations to set up temporary accommodation until their homes are inhabitable again. A particular challenge is understanding how many vulnerable people, such as children and the elderly, may be in each of these new locations.
In a research project by the Imperial College London, AI has been used to determine the gender of mobile phone users to help in such situations. Mobile phone use has increased rapidly in developing countries in recent years. However, due to the large number of pre-paid contracts, mobile phone service providers have very little data on their subscribers, data that could help in a crisis. By analyzing the data of 10,000 users using AI, it was possible to reliably predict the gender of a user by the way they used their mobile phone.
With an accuracy of around 80%, some 10% better than previous studies, determining the gender-balance of a group would help to prioritize the type of aid needed by a displaced group. In this way, AI could help to complement mobile phone data which is already used to determine people’s locations.
The UN’s World Food Programme (WFP) is also reviewing how AI could support their disaster relief efforts. It could be used to quickly review drone footage and satellite imagery to support planning efforts. Unmanned Aerial Vehicles (UAV) could also make use of AI to deliver much-needed supplies to remote regions unreachable by manned vehicles.
But could AI even be used to predict the events that lead to mass-migration, such as conflict, famine and even economic changes? Babusi Nyoni, someone who migrated from Zimbabwe to South Africa for socio-economic reasons, has been exploring whether AI could anticipate such occurrences. Using data from the world bank, information such as Gross Domestic Product (GDP) and Food Production Index, coupled with climate change data and political announcements, he feels that this could be possible.
Natural disaster, conflict and famine can have an impact that lasts a generation in the developing world. By applying AI tools and concepts to these challenges, it could see huge improvements in the success of humanitarian response in the face of such disasters.